diff --git a/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_A100_ppl_30_0797_20260417_105249/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_A100_ppl_30_0797_20260417_105249.txt b/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_A100_ppl_30_0797_20260417_105249/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_A100_ppl_30_0797_20260417_105249.txt new file mode 100644 index 0000000000000000000000000000000000000000..f65319967d0ee378732bce4c9f0c21b62d214aba --- /dev/null +++ b/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_A100_ppl_30_0797_20260417_105249/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_A100_ppl_30_0797_20260417_105249.txt @@ -0,0 +1,220 @@ +W0417 10:52:53.221406 112943 site-packages/torch/distributed/run.py:793] +W0417 10:52:53.221406 112943 site-packages/torch/distributed/run.py:793] ***************************************** +W0417 10:52:53.221406 112943 site-packages/torch/distributed/run.py:793] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0417 10:52:53.221406 112943 site-packages/torch/distributed/run.py:793] ***************************************** +Starting script +Starting script +Starting script +Starting script +2026-04-17 10:53:14.383 | INFO | __main__:main:45 - Global rank 0, local rank 0, device: 0 +[W417 10:53:14.717367333 CUDAAllocatorConfig.h:28] Warning: expandable_segments not supported on this platform (function operator()) +2026-04-17 10:53:14.388 | INFO | __main__:main:48 - Process group initialized +2026-04-17 10:53:14.388 | INFO | __main__:main:54 - Using dist with rank 0 (only rank 0 will log) +2026-04-17 10:53:14.388 | INFO | __main__:main:55 - **************************************** +2026-04-17 10:53:14.388 | INFO | __main__:main:56 - Starting training with the arguments +2026-04-17 10:53:14.388 | INFO | __main__:main:58 - model_config configs/llama_60m.json +2026-04-17 10:53:14.388 | INFO | __main__:main:58 - exp_config exp_v2/configs/llama_60m.json +2026-04-17 10:53:14.388 | INFO | __main__:main:58 - eval_every 1000 +2026-04-17 10:53:14.388 | INFO | __main__:main:58 - save_every 10000 +2026-04-17 10:53:14.388 | INFO | __main__:main:58 - dtype bfloat16 +2026-04-17 10:53:14.388 | INFO | __main__:main:58 - seed 0 +2026-04-17 10:53:14.388 | INFO | __main__:main:58 - compile True +2026-04-17 10:53:14.388 | INFO | __main__:main:58 - dynamo_suppress_errors True +2026-04-17 10:53:14.388 | INFO | __main__:main:58 - dynamo_cache_limit 10000 +2026-04-17 10:53:14.388 | INFO | __main__:main:58 - memory_cleanup_frequency 10000 +2026-04-17 10:53:14.388 | INFO | __main__:main:58 - resume_step None +2026-04-17 10:53:14.388 | INFO | __main__:main:58 - restore_optimizer False +2026-04-17 10:53:14.388 | INFO | __main__:main:58 - continue_from None +2026-04-17 10:53:14.388 | INFO | __main__:main:58 - single_gpu False +2026-04-17 10:53:14.388 | INFO | __main__:main:58 - save_dir /mnt/dhwfile/tancheng/panjiabao/Result/SAC_C4/work_dirs/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_20260417_105249 +2026-04-17 10:53:14.388 | INFO | __main__:main:58 - use_hf_model False +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - workers 12 +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - batch_size 128 +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - gradient_accumulation None +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - total_batch_size 512 +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - lr 0.003 +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - warmup_steps 1000 +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - num_training_steps 10000 +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - max_train_tokens None +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - optimizer adamw +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - max_length 256 +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - scheduler cosine +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - min_lr_ratio 0.1 +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - weight_decay 0.0 +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - grad_clipping 0.0 +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - activation_checkpointing False +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - data_path /mnt/dhwfile/tancheng/panjiabao/dataset/C4/en +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - data_name en +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - tags None +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - name test +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - project test +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - unset_wandb False +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - entity None +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - wandb_dir /mnt/dhwfile/tancheng/panjiabao/Result/SAC_C4/work_dirs/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_20260417_105249 +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - beta1 0.9 +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - beta2 0.99 +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - beta3 0.99 +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - eps 1e-08 +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - rank 128 +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - update_proj_gap 50 +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - galore_scale 1.0 +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - proj_type std +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - proj_quant False +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - proj_bits 8 +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - proj_group_size 256 +2026-04-17 10:53:14.389 | INFO | __main__:main:58 - weight_quant False +2026-04-17 10:53:14.390 | INFO | __main__:main:58 - weight_bits 8 +2026-04-17 10:53:14.390 | INFO | __main__:main:58 - weight_group_size 256 +2026-04-17 10:53:14.390 | INFO | __main__:main:58 - stochastic_round False +2026-04-17 10:53:14.390 | INFO | __main__:main:58 - simulation False +2026-04-17 10:53:14.390 | INFO | __main__:main:58 - cos_threshold 1 +2026-04-17 10:53:14.390 | INFO | __main__:main:58 - gamma_proj 2 +2026-04-17 10:53:14.390 | INFO | __main__:main:58 - queue_size 5 +2026-04-17 10:53:14.390 | INFO | __main__:main:58 - proj random +2026-04-17 10:53:14.390 | INFO | __main__:main:58 - scale_type tensor +2026-04-17 10:53:14.390 | INFO | __main__:main:58 - apollo_scale 1.0 +2026-04-17 10:53:14.390 | INFO | __main__:main:58 - scale_front False +2026-04-17 10:53:14.390 | INFO | __main__:main:58 - n_clusters 3 +2026-04-17 10:53:14.390 | INFO | __main__:main:58 - scale_update_freq 500 +2026-04-17 10:53:14.390 | INFO | __main__:main:58 - scale_level 1,0,1,1 +2026-04-17 10:53:14.390 | INFO | __main__:main:58 - scale_bound None +2026-04-17 10:53:14.390 | INFO | __main__:main:58 - metric mean +2026-04-17 10:53:14.390 | INFO | __main__:main:58 - align_grad False +2026-04-17 10:53:14.390 | INFO | __main__:main:58 - dim 4096 +2026-04-17 10:53:14.390 | INFO | __main__:main:58 - n_heads 32 +2026-04-17 10:53:14.390 | INFO | __main__:main:58 - muon_ns_steps 5 +2026-04-17 10:53:14.390 | INFO | __main__:main:58 - muon_momentum 0.95 +2026-04-17 10:53:14.390 | INFO | __main__:main:58 - nproc_per_node 4 +2026-04-17 10:53:14.390 | INFO | __main__:main:59 - **************************************** +2026-04-17 10:53:14.390 | INFO | __main__:main:68 - Wandb directory set to: /mnt/dhwfile/tancheng/panjiabao/Result/SAC_C4/work_dirs/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_20260417_105249 +2026-04-17 10:53:14.600 | INFO | __main__:main:45 - Global rank 3, local rank 3, device: 3 +2026-04-17 10:53:14.600 | INFO | __main__:main:45 - Global rank 2, local rank 2, device: 2 +2026-04-17 10:53:14.601 | INFO | __main__:main:45 - Global rank 1, local rank 1, device: 1 +[W417 10:53:14.932430355 CUDAAllocatorConfig.h:28] Warning: expandable_segments not supported on this platform (function operator()) +[W417 10:53:14.932467426 CUDAAllocatorConfig.h:28] Warning: expandable_segments not supported on this platform (function operator()) +[W417 10:53:14.932471703 CUDAAllocatorConfig.h:28] Warning: expandable_segments not supported on this platform (function operator()) +2026-04-17 10:53:14.603 | INFO | __main__:main:48 - Process group initialized +2026-04-17 10:53:14.603 | INFO | __main__:main:48 - Process group initialized +2026-04-17 10:53:14.603 | INFO | __main__:main:48 - Process group initialized +wandb: Tracking run with wandb version 0.23.0 +wandb: W&B syncing is set to `offline` in this directory. Run `wandb online` or set WANDB_MODE=online to enable cloud syncing. +wandb: Run data is saved locally in /mnt/dhwfile/tancheng/panjiabao/Result/SAC_C4/work_dirs/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_20260417_105249/wandb/offline-run-20260417_105314-6uk0fpmn + Resolving data files: 0%| | 0/1032 [00:00 will be ignored +[rank2]:W0417 10:53:22.330679 113017 site-packages/torch/_logging/_internal.py:1081] [0/0] Profiler function will be ignored +[rank3]:W0417 10:53:22.354559 113019 site-packages/torch/_logging/_internal.py:1081] [0/0] Profiler function will be ignored +[rank1]:W0417 10:53:22.404604 113016 site-packages/torch/_logging/_internal.py:1081] [0/0] Profiler function will be ignored + Update steps: 0%| | 1/10000 [00:17<48:05:06, 17.31s/it] Update steps: 0%| | 2/10000 [00:17<20:01:12, 7.21s/it] Update steps: 0%| | 4/10000 [00:17<7:35:09, 2.73s/it] Update steps: 0%| | 6/10000 [00:17<4:05:38, 1.47s/it] Update steps: 0%| | 8/10000 [00:17<2:31:35, 1.10it/s] Update steps: 0%| | 10/10000 [00:18<1:40:24, 1.66it/s] Update steps: 0%| | 12/10000 [00:18<1:10:12, 2.37it/s] Update steps: 0%| | 14/10000 [00:18<51:11, 3.25it/s] Update steps: 0%| | 16/10000 [00:18<38:35, 4.31it/s] Update steps: 0%| | 18/10000 [00:18<30:14, 5.50it/s] Update steps: 0%| | 20/10000 [00:18<24:34, 6.77it/s] Update steps: 0%| | 22/10000 [00:18<20:31, 8.10it/s] Update steps: 0%| | 24/10000 [00:19<17:56, 9.27it/s] Update steps: 0%| | 26/10000 [00:19<16:04, 10.34it/s] Update steps: 0%| | 28/10000 [00:19<14:40, 11.32it/s] Update steps: 0%| | 30/10000 [00:19<13:42, 12.11it/s] Update steps: 0%| | 32/10000 [00:19<13:05, 12.69it/s] Update steps: 0%| | 34/10000 [00:19<12:44, 13.03it/s] Update steps: 0%| | 36/10000 [00:19<12:29, 13.30it/s] Update steps: 0%| | 38/10000 [00:20<12:23, 13.40it/s] Update steps: 0%| | 40/10000 [00:20<12:09, 13.65it/s] Update steps: 0%| | 42/10000 [00:20<12:03, 13.77it/s] Update steps: 0%| | 44/10000 [00:20<11:53, 13.94it/s] Update steps: 0%| | 46/10000 [00:20<11:54, 13.94it/s] Update steps: 0%| | 48/10000 [00:20<11:46, 14.09it/s] Update steps: 0%|▏ | 50/10000 [00:20<11:40, 14.20it/s] Update steps: 1%|▏ | 52/10000 [00:20<11:39, 14.22it/s] Update steps: 1%|▏ | 54/10000 [00:21<11:38, 14.25it/s] Update steps: 1%|▏ | 56/10000 [00:21<11:36, 14.28it/s] Update steps: 1%|▏ | 58/10000 [00:21<11:40, 14.20it/s] Update steps: 1%|▏ | 60/10000 [00:21<11:47, 14.05it/s] Update steps: 1%|▏ | 62/10000 [00:21<11:40, 14.18it/s] Update steps: 1%|▏ | 64/10000 [00:21<11:42, 14.15it/s] Update steps: 1%|▏ | 66/10000 [00:21<11:47, 14.05it/s] Update steps: 1%|▏ | 68/10000 [00:22<11:46, 14.05it/s] Update steps: 1%|▏ | 70/10000 [00:22<11:44, 14.10it/s] Update steps: 1%|▏ | 72/10000 [00:22<11:38, 14.22it/s] Update steps: 1%|▏ | 74/10000 [00:22<11:39, 14.18it/s] Update steps: 1%|▏ | 76/10000 [00:22<11:42, 14.13it/s] Update steps: 1%|▏ | 78/10000 [00:22<11:47, 14.03it/s] Update steps: 1%|▏ | 80/10000 [00:22<11:45, 14.07it/s] Update steps: 1%|▏ | 82/10000 [00:23<11:38, 14.19it/s] Update steps: 1%|▏ | 84/10000 [00:23<11:36, 14.24it/s] Update steps: 1%|▏ | 86/10000 [00:23<11:38, 14.19it/s] Update steps: 1%|▏ | 88/10000 [00:23<11:36, 14.23it/s] Update steps: 1%|▏ | 90/10000 [00:23<11:40, 14.15it/s] Update steps: 1%|▏ | 92/10000 [00:23<11:36, 14.23it/s] Update steps: 1%|▏ | 94/10000 [00:23<11:37, 14.21it/s] Update steps: 1%|▏ | 96/10000 [00:24<11:40, 14.13it/s] Update steps: 1%|▎ | 98/10000 [00:24<11:40, 14.14it/s] Update steps: 1%|▎ | 100/10000 [00:24<11:39, 14.16it/s] Update steps: 1%|▎ | 102/10000 [00:24<11:42, 14.10it/s] Update steps: 1%|▎ | 104/10000 [00:24<11:46, 14.02it/s] Update steps: 1%|▎ | 106/10000 [00:24<11:49, 13.95it/s] Update steps: 1%|▎ | 108/10000 [00:24<11:45, 14.03it/s] Update steps: 1%|▎ | 110/10000 [00:25<11:52, 13.88it/s] Update steps: 1%|▎ | 112/10000 [00:25<11:46, 14.01it/s] Update steps: 1%|▎ | 114/10000 [00:25<11:54, 13.84it/s] Update steps: 1%|▎ | 116/10000 [00:25<11:47, 13.96it/s] Update steps: 1%|▎ | 118/10000 [00:25<11:47, 13.98it/s] Update steps: 1%|▎ | 120/10000 [00:25<11:45, 14.00it/s] Update steps: 1%|▎ | 122/10000 [00:25<11:45, 13.99it/s] Update steps: 1%|▎ | 124/10000 [00:26<11:46, 13.98it/s] Update steps: 1%|▎ | 126/10000 [00:26<11:42, 14.06it/s] Update steps: 1%|▎ | 128/10000 [00:26<11:43, 14.02it/s] Update steps: 1%|▎ | 130/10000 [00:26<11:41, 14.08it/s] Update steps: 1%|▎ | 132/10000 [00:26<11:43, 14.02it/s] Update steps: 1%|▎ | 134/10000 [00:26<11:37, 14.15it/s] Update steps: 1%|▎ | 136/10000 [00:26<11:33, 14.21it/s] Update steps: 1%|▎ | 138/10000 [00:27<11:38, 14.11it/s] Update steps: 1%|▎ | 140/10000 [00:27<11:40, 14.07it/s] Update steps: 1%|▎ | 142/10000 [00:27<11:49, 13.89it/s] Update steps: 1%|▎ | 144/10000 [00:27<11:48, 13.92it/s] Update steps: 1%|▎ | 146/10000 [00:27<11:44, 13.99it/s] Update steps: 1%|▎ | 148/10000 [00:27<11:46, 13.95it/s] Update steps: 2%|▍ | 150/10000 [00:27<11:44, 13.98it/s] Update steps: 2%|▍ | 152/10000 [00:28<11:37, 14.12it/s] Update steps: 2%|▍ | 154/10000 [00:28<11:41, 14.03it/s] Update steps: 2%|▍ | 156/10000 [00:28<11:42, 14.00it/s] Update steps: 2%|▍ | 158/10000 [00:28<11:41, 14.03it/s] Update steps: 2%|▍ | 160/10000 [00:28<11:43, 13.99it/s] Update steps: 2%|▍ | 162/10000 [00:28<11:39, 14.07it/s] Update steps: 2%|▍ | 164/10000 [00:28<11:41, 14.03it/s] Update steps: 2%|▍ | 166/10000 [00:29<11:46, 13.93it/s] Update steps: 2%|▍ | 168/10000 [00:29<11:41, 14.02it/s] Update steps: 2%|▍ | 170/10000 [00:29<11:38, 14.07it/s] Update steps: 2%|▍ | 172/10000 [00:29<11:33, 14.16it/s] Update steps: 2%|▍ | 174/10000 [00:29<11:31, 14.21it/s] Update steps: 2%|▍ | 176/10000 [00:29<11:29, 14.25it/s] Update steps: 2%|▍ | 178/10000 [00:29<11:37, 14.08it/s] Update steps: 2%|▍ | 180/10000 [00:30<11:43, 13.96it/s] Update steps: 2%|▍ | 182/10000 [00:30<11:46, 13.89it/s] Update steps: 2%|▍ | 184/10000 [00:30<11:49, 13.83it/s] Update steps: 2%|▍ | 186/10000 [00:30<11:44, 13.93it/s] Update steps: 2%|▍ | 188/10000 [00:30<11:45, 13.92it/s] Update steps: 2%|▍ | 190/10000 [00:30<11:38, 14.05it/s] Update steps: 2%|▍ | 192/10000 [00:30<11:34, 14.12it/s] Update steps: 2%|▍ | 194/10000 [00:31<11:31, 14.17it/s] Update steps: 2%|▍ | 196/10000 [00:31<11:33, 14.13it/s] Update steps: 2%|▍ | 198/10000 [00:31<11:32, 14.15it/s] Update steps: 2%|▌ | 200/10000 [00:31<11:35, 14.09it/s] Update steps: 2%|▌ | 202/10000 [00:31<11:39, 14.00it/s] Update steps: 2%|▌ | 204/10000 [00:31<11:39, 14.00it/s] Update steps: 2%|▌ | 206/10000 [00:31<11:36, 14.06it/s] Update steps: 2%|▌ | 208/10000 [00:32<11:34, 14.11it/s] Update steps: 2%|▌ | 210/10000 [00:32<11:29, 14.19it/s] Update steps: 2%|▌ | 212/10000 [00:32<11:37, 14.04it/s] Update steps: 2%|▌ | 214/10000 [00:32<11:45, 13.88it/s] Update steps: 2%|▌ | 216/10000 [00:32<11:43, 13.91it/s] Update steps: 2%|▌ | 218/10000 [00:32<11:42, 13.92it/s] Update steps: 2%|▌ | 220/10000 [00:32<11:45, 13.86it/s] Update steps: 2%|▌ | 222/10000 [00:33<11:43, 13.89it/s] Update steps: 2%|▌ | 224/10000 [00:33<11:37, 14.02it/s] Update steps: 2%|▌ | 226/10000 [00:33<11:34, 14.07it/s] Update steps: 2%|▌ | 228/10000 [00:33<11:30, 14.15it/s] Update steps: 2%|▌ | 230/10000 [00:33<11:27, 14.20it/s] Update steps: 2%|▌ | 232/10000 [00:33<11:27, 14.20it/s] Update steps: 2%|▌ | 234/10000 [00:33<11:28, 14.18it/s] Update steps: 2%|▌ | 236/10000 [00:34<11:32, 14.11it/s] Update steps: 2%|▌ | 238/10000 [00:34<11:32, 14.10it/s] Update steps: 2%|▌ | 240/10000 [00:34<11:37, 13.99it/s] Update steps: 2%|▌ | 242/10000 [00:34<11:38, 13.97it/s] Update steps: 2%|▌ | 244/10000 [00:34<11:41, 13.92it/s] Update steps: 2%|▌ | 246/10000 [00:34<11:42, 13.88it/s] Update steps: 2%|▌ | 248/10000 [00:34<11:46, 13.80it/s] Update steps: 2%|▋ | 250/10000 [00:35<11:43, 13.86it/s] Update steps: 3%|▋ | 252/10000 [00:35<11:41, 13.91it/s] Update steps: 3%|▋ | 254/10000 [00:35<11:44, 13.82it/s] Update steps: 3%|▋ | 256/10000 [00:35<11:44, 13.84it/s] Update steps: 3%|▋ | 258/10000 [00:35<11:44, 13.83it/s] Update steps: 3%|▋ | 260/10000 [00:35<11:38, 13.95it/s] Update steps: 3%|▋ | 262/10000 [00:35<11:33, 14.05it/s] Update steps: 3%|▋ | 264/10000 [00:36<11:34, 14.02it/s] Update steps: 3%|▋ | 266/10000 [00:36<11:38, 13.93it/s] Update steps: 3%|▋ | 268/10000 [00:36<11:32, 14.05it/s] Update steps: 3%|▋ | 270/10000 [00:36<11:35, 13.99it/s] Update steps: 3%|▋ | 272/10000 [00:36<11:39, 13.92it/s] Update steps: 3%|▋ | 274/10000 [00:36<11:41, 13.87it/s] Update steps: 3%|▋ | 276/10000 [00:36<11:34, 14.00it/s] Update steps: 3%|▋ | 278/10000 [00:37<11:29, 14.11it/s] Update steps: 3%|▋ | 280/10000 [00:37<11:30, 14.09it/s] Update steps: 3%|▋ | 282/10000 [00:37<11:27, 14.13it/s] Update steps: 3%|▋ | 284/10000 [00:37<11:30, 14.07it/s] Update steps: 3%|▋ | 286/10000 [00:37<11:40, 13.88it/s] Update steps: 3%|▋ | 288/10000 [00:37<11:39, 13.88it/s] Update steps: 3%|▋ | 290/10000 [00:37<11:45, 13.76it/s] Update steps: 3%|▋ | 292/10000 [00:38<11:44, 13.78it/s] Update steps: 3%|▋ | 294/10000 [00:38<11:43, 13.80it/s] Update steps: 3%|▋ | 296/10000 [00:38<11:42, 13.82it/s] Update steps: 3%|▋ | 298/10000 [00:38<11:48, 13.70it/s] Update steps: 3%|▊ | 300/10000 [00:38<11:44, 13.77it/s] Update steps: 3%|▊ | 302/10000 [00:38<11:36, 13.91it/s] Update steps: 3%|▊ | 304/10000 [00:38<11:40, 13.84it/s] Update steps: 3%|▊ | 306/10000 [00:39<11:34, 13.97it/s] Update steps: 3%|▊ | 308/10000 [00:39<11:38, 13.87it/s] Update steps: 3%|▊ | 310/10000 [00:39<11:45, 13.74it/s] Update steps: 3%|▊ | 312/10000 [00:39<11:36, 13.91it/s] Update steps: 3%|▊ | 314/10000 [00:39<11:36, 13.92it/s] Update steps: 3%|▊ | 316/10000 [00:39<11:32, 13.99it/s] Update steps: 3%|▊ | 318/10000 [00:39<11:37, 13.88it/s] Update steps: 3%|▊ | 320/10000 [00:40<11:38, 13.86it/s] Update steps: 3%|▊ | 322/10000 [00:40<11:42, 13.78it/s] Update steps: 3%|▊ | 324/10000 [00:40<11:43, 13.75it/s] Update steps: 3%|▊ | 326/10000 [00:40<11:41, 13.79it/s] Update steps: 3%|▊ | 328/10000 [00:40<11:47, 13.66it/s] Update steps: 3%|▊ | 330/10000 [00:40<11:50, 13.62it/s] Update steps: 3%|▊ | 332/10000 [00:41<11:40, 13.80it/s] Update steps: 3%|▊ | 334/10000 [00:41<11:37, 13.86it/s] Update steps: 3%|▊ | 336/10000 [00:41<11:34, 13.91it/s] Update steps: 3%|▊ | 338/10000 [00:41<11:36, 13.88it/s] Update steps: 3%|▊ | 340/10000 [00:41<11:34, 13.91it/s] Update steps: 3%|▊ | 342/10000 [00:41<11:29, 14.01it/s] Update steps: 3%|▊ | 344/10000 [00:41<11:25, 14.09it/s] Update steps: 3%|▊ | 346/10000 [00:42<11:30, 13.99it/s] Update steps: 3%|▊ | 348/10000 [00:42<11:29, 13.99it/s] Update steps: 4%|▉ | 350/10000 [00:42<11:28, 14.01it/s] Update steps: 4%|▉ | 352/10000 [00:42<11:26, 14.06it/s] Update steps: 4%|▉ | 354/10000 [00:42<11:24, 14.10it/s] Update steps: 4%|▉ | 356/10000 [00:42<11:36, 13.85it/s] Update steps: 4%|▉ | 358/10000 [00:42<11:37, 13.82it/s] Update steps: 4%|▉ | 360/10000 [00:43<11:37, 13.82it/s] Update steps: 4%|▉ | 362/10000 [00:43<11:37, 13.82it/s] Update steps: 4%|▉ | 364/10000 [00:43<11:30, 13.95it/s] Update steps: 4%|▉ | 366/10000 [00:43<11:34, 13.86it/s] Update steps: 4%|▉ | 368/10000 [00:43<11:29, 13.97it/s] Update steps: 4%|▉ | 370/10000 [00:43<11:31, 13.93it/s] Update steps: 4%|▉ | 372/10000 [00:43<11:32, 13.90it/s] Update steps: 4%|▉ | 374/10000 [00:44<11:26, 14.02it/s] Update steps: 4%|▉ | 376/10000 [00:44<11:30, 13.94it/s] Update steps: 4%|▉ | 378/10000 [00:44<11:25, 14.04it/s] Update steps: 4%|▉ | 380/10000 [00:44<11:31, 13.92it/s] Update steps: 4%|▉ | 382/10000 [00:44<11:34, 13.85it/s] Update steps: 4%|▉ | 384/10000 [00:44<11:38, 13.76it/s] Update steps: 4%|▉ | 386/10000 [00:44<11:36, 13.81it/s] Update steps: 4%|▉ | 388/10000 [00:45<11:41, 13.71it/s] Update steps: 4%|▉ | 390/10000 [00:45<11:41, 13.70it/s] Update steps: 4%|▉ | 392/10000 [00:45<11:33, 13.86it/s] Update steps: 4%|▉ | 394/10000 [00:45<11:33, 13.86it/s] Update steps: 4%|▉ | 396/10000 [00:45<11:34, 13.84it/s] Update steps: 4%|▉ | 398/10000 [00:45<11:33, 13.84it/s] Update steps: 4%|█ | 400/10000 [00:45<11:39, 13.72it/s] Update steps: 4%|█ | 402/10000 [00:46<11:39, 13.73it/s] Update steps: 4%|█ | 404/10000 [00:46<11:36, 13.77it/s] Update steps: 4%|█ | 406/10000 [00:46<11:36, 13.78it/s] Update steps: 4%|█ | 408/10000 [00:46<11:29, 13.90it/s] Update steps: 4%|█ | 410/10000 [00:46<11:31, 13.87it/s] Update steps: 4%|█ | 412/10000 [00:46<11:39, 13.72it/s] Update steps: 4%|█ | 414/10000 [00:46<11:31, 13.87it/s] Update steps: 4%|█ | 416/10000 [00:47<11:27, 13.93it/s] Update steps: 4%|█ | 418/10000 [00:47<11:26, 13.96it/s] Update steps: 4%|█ | 420/10000 [00:47<11:32, 13.84it/s] Update steps: 4%|█ | 422/10000 [00:47<11:24, 13.99it/s] Update steps: 4%|█ | 424/10000 [00:47<11:21, 14.05it/s] Update steps: 4%|█ | 426/10000 [00:47<11:26, 13.95it/s] Update steps: 4%|█ | 428/10000 [00:47<11:34, 13.78it/s] Update steps: 4%|█ | 430/10000 [00:48<11:40, 13.67it/s] Update steps: 4%|█ | 432/10000 [00:48<11:39, 13.68it/s] Update steps: 4%|█ | 434/10000 [00:48<11:43, 13.59it/s] Update steps: 4%|█ | 436/10000 [00:48<11:37, 13.72it/s] Update steps: 4%|█ | 438/10000 [00:48<11:37, 13.72it/s] Update steps: 4%|█ | 440/10000 [00:48<11:45, 13.55it/s] Update steps: 4%|█ | 442/10000 [00:48<11:40, 13.64it/s] Update steps: 4%|█ | 444/10000 [00:49<11:34, 13.76it/s] Update steps: 4%|█ | 446/10000 [00:49<11:35, 13.74it/s] Update steps: 4%|█ | 448/10000 [00:49<11:36, 13.71it/s] Update steps: 4%|█▏ | 450/10000 [00:49<11:41, 13.62it/s] Update steps: 5%|█▏ | 452/10000 [00:49<11:43, 13.58it/s] Update steps: 5%|█▏ | 454/10000 [00:49<11:34, 13.75it/s] Update steps: 5%|█▏ | 456/10000 [00:49<11:33, 13.76it/s] Update steps: 5%|█▏ | 458/10000 [00:50<11:35, 13.71it/s] Update steps: 5%|█▏ | 460/10000 [00:50<11:34, 13.74it/s] Update steps: 5%|█▏ | 462/10000 [00:50<11:27, 13.87it/s] Update steps: 5%|█▏ | 464/10000 [00:50<11:27, 13.87it/s] Update steps: 5%|█▏ | 466/10000 [00:50<11:34, 13.73it/s] Update steps: 5%|█▏ | 468/10000 [00:50<11:29, 13.82it/s] Update steps: 5%|█▏ | 470/10000 [00:50<11:30, 13.79it/s] Update steps: 5%|█▏ | 472/10000 [00:51<11:32, 13.75it/s] Update steps: 5%|█▏ | 474/10000 [00:51<11:39, 13.62it/s] Update steps: 5%|█▏ | 476/10000 [00:51<11:40, 13.60it/s] Update steps: 5%|█▏ | 478/10000 [00:51<11:33, 13.73it/s] Update steps: 5%|█▏ | 480/10000 [00:51<11:32, 13.74it/s] Update steps: 5%|█▏ | 482/10000 [00:51<11:26, 13.87it/s] Update steps: 5%|█▏ | 484/10000 [00:51<11:26, 13.86it/s] Update steps: 5%|█▏ | 486/10000 [00:52<11:30, 13.77it/s] Update steps: 5%|█▏ | 488/10000 [00:52<11:29, 13.81it/s] Update steps: 5%|█▏ | 490/10000 [00:52<11:24, 13.89it/s] Update steps: 5%|█▏ | 492/10000 [00:52<11:27, 13.82it/s] Update steps: 5%|█▏ | 494/10000 [00:52<11:32, 13.72it/s] Update steps: 5%|█▏ | 496/10000 [00:52<11:29, 13.79it/s] Update steps: 5%|█▏ | 498/10000 [00:53<11:25, 13.85it/s] Update steps: 5%|█▎ | 500/10000 [00:53<11:27, 13.82it/s] Update steps: 5%|█▎ | 502/10000 [00:53<11:27, 13.81it/s] Update steps: 5%|█▎ | 504/10000 [00:53<11:28, 13.80it/s] Update steps: 5%|█▎ | 506/10000 [00:53<11:27, 13.82it/s] Update steps: 5%|█▎ | 508/10000 [00:53<11:24, 13.87it/s] Update steps: 5%|█▎ | 510/10000 [00:53<11:23, 13.88it/s] Update steps: 5%|█▎ | 512/10000 [00:54<11:18, 13.98it/s] Update steps: 5%|█▎ | 514/10000 [00:54<11:16, 14.03it/s] Update steps: 5%|█▎ | 516/10000 [00:54<11:18, 13.97it/s] Update steps: 5%|█▎ | 518/10000 [00:54<11:21, 13.90it/s] Update steps: 5%|█▎ | 520/10000 [00:54<11:19, 13.95it/s] Update steps: 5%|█▎ | 522/10000 [00:54<11:17, 13.99it/s] Update steps: 5%|█▎ | 524/10000 [00:54<11:14, 14.04it/s] Update steps: 5%|█▎ | 526/10000 [00:55<11:13, 14.07it/s] Update steps: 5%|█▎ | 528/10000 [00:55<11:12, 14.07it/s] Update steps: 5%|█▎ | 530/10000 [00:55<11:15, 14.03it/s] Update steps: 5%|█▎ | 532/10000 [00:55<11:18, 13.95it/s] Update steps: 5%|█▎ | 534/10000 [00:55<11:17, 13.97it/s] Update steps: 5%|█▎ | 536/10000 [00:55<11:24, 13.83it/s] Update steps: 5%|█▎ | 538/10000 [00:55<11:31, 13.69it/s] Update steps: 5%|█▎ | 540/10000 [00:56<11:32, 13.66it/s] Update steps: 5%|█▎ | 542/10000 [00:56<11:26, 13.78it/s] Update steps: 5%|█▎ | 544/10000 [00:56<11:24, 13.82it/s] Update steps: 5%|█▎ | 546/10000 [00:56<11:22, 13.85it/s] Update steps: 5%|█▎ | 548/10000 [00:56<11:26, 13.77it/s] Update steps: 6%|█▍ | 550/10000 [00:56<11:24, 13.80it/s] Update steps: 6%|█▍ | 552/10000 [00:56<11:21, 13.87it/s] Update steps: 6%|█▍ | 554/10000 [00:57<11:16, 13.96it/s] Update steps: 6%|█▍ | 556/10000 [00:57<11:20, 13.88it/s] Update steps: 6%|█▍ | 558/10000 [00:57<11:20, 13.88it/s] Update steps: 6%|█▍ | 560/10000 [00:57<11:26, 13.76it/s] Update steps: 6%|█▍ | 562/10000 [00:57<11:25, 13.76it/s] Update steps: 6%|█▍ | 564/10000 [00:57<11:23, 13.80it/s] Update steps: 6%|█▍ | 566/10000 [00:57<11:22, 13.83it/s] Update steps: 6%|█▍ | 568/10000 [00:58<11:19, 13.88it/s] Update steps: 6%|█▍ | 570/10000 [00:58<11:17, 13.91it/s] Update steps: 6%|█▍ | 572/10000 [00:58<11:21, 13.83it/s] Update steps: 6%|█▍ | 574/10000 [00:58<11:26, 13.74it/s] Update steps: 6%|█▍ | 576/10000 [00:58<11:26, 13.73it/s] Update steps: 6%|█▍ | 578/10000 [00:58<11:28, 13.69it/s] Update steps: 6%|█▍ | 580/10000 [00:58<11:24, 13.76it/s] Update steps: 6%|█▍ | 582/10000 [00:59<11:20, 13.84it/s] Update steps: 6%|█▍ | 584/10000 [00:59<11:17, 13.90it/s] Update steps: 6%|█▍ | 586/10000 [00:59<11:12, 14.00it/s] Update steps: 6%|█▍ | 588/10000 [00:59<11:16, 13.90it/s] Update steps: 6%|█▍ | 590/10000 [00:59<11:19, 13.84it/s] Update steps: 6%|█▍ | 592/10000 [00:59<11:28, 13.65it/s] Update steps: 6%|█▍ | 594/10000 [00:59<11:26, 13.71it/s] Update steps: 6%|█▍ | 596/10000 [01:00<11:28, 13.66it/s] Update steps: 6%|█▍ | 598/10000 [01:00<11:30, 13.62it/s] Update steps: 6%|█▌ | 600/10000 [01:00<11:28, 13.65it/s] Update steps: 6%|█▌ | 602/10000 [01:00<11:24, 13.74it/s] Update steps: 6%|█▌ | 604/10000 [01:00<11:19, 13.84it/s] Update steps: 6%|█▌ | 606/10000 [01:00<11:14, 13.93it/s] Update steps: 6%|█▌ | 608/10000 [01:00<11:15, 13.91it/s] Update steps: 6%|█▌ | 610/10000 [01:01<11:12, 13.96it/s] Update steps: 6%|█▌ | 612/10000 [01:01<11:16, 13.88it/s] Update steps: 6%|█▌ | 614/10000 [01:01<11:18, 13.83it/s] Update steps: 6%|█▌ | 616/10000 [01:01<11:21, 13.76it/s] Update steps: 6%|█▌ | 618/10000 [01:01<11:18, 13.82it/s] Update steps: 6%|█▌ | 620/10000 [01:01<11:17, 13.84it/s] Update steps: 6%|█▌ | 622/10000 [01:01<11:20, 13.78it/s] Update steps: 6%|█▌ | 624/10000 [01:02<11:23, 13.71it/s] Update steps: 6%|█▌ | 626/10000 [01:02<11:28, 13.61it/s] Update steps: 6%|█▌ | 628/10000 [01:02<11:29, 13.60it/s] Update steps: 6%|█▌ | 630/10000 [01:02<11:24, 13.69it/s] Update steps: 6%|█▌ | 632/10000 [01:02<11:18, 13.81it/s] Update steps: 6%|█▌ | 634/10000 [01:02<11:13, 13.92it/s] Update steps: 6%|█▌ | 636/10000 [01:02<11:08, 14.00it/s] Update steps: 6%|█▌ | 638/10000 [01:03<11:10, 13.96it/s] Update steps: 6%|█▌ | 640/10000 [01:03<11:09, 13.97it/s] Update steps: 6%|█▌ | 642/10000 [01:03<11:14, 13.88it/s] Update steps: 6%|█▌ | 644/10000 [01:03<11:13, 13.89it/s] Update steps: 6%|█▌ | 646/10000 [01:03<11:13, 13.89it/s] Update steps: 6%|█▌ | 648/10000 [01:03<11:11, 13.92it/s] Update steps: 6%|█▋ | 650/10000 [01:03<11:11, 13.92it/s] Update steps: 7%|█▋ | 652/10000 [01:04<11:14, 13.87it/s] Update steps: 7%|█▋ | 654/10000 [01:04<11:17, 13.79it/s] Update steps: 7%|█▋ | 656/10000 [01:04<11:19, 13.74it/s] Update steps: 7%|█▋ | 658/10000 [01:04<11:18, 13.76it/s] Update steps: 7%|█▋ | 660/10000 [01:04<11:24, 13.64it/s] Update steps: 7%|█▋ | 662/10000 [01:04<11:23, 13.66it/s] Update steps: 7%|█▋ | 664/10000 [01:05<11:27, 13.58it/s] Update steps: 7%|█▋ | 666/10000 [01:05<11:24, 13.63it/s] Update steps: 7%|█▋ | 668/10000 [01:05<11:20, 13.72it/s] Update steps: 7%|█▋ | 670/10000 [01:05<11:17, 13.77it/s] Update steps: 7%|█▋ | 672/10000 [01:05<11:21, 13.68it/s] Update steps: 7%|█▋ | 674/10000 [01:05<11:17, 13.76it/s] Update steps: 7%|█▋ | 676/10000 [01:05<11:16, 13.78it/s] Update steps: 7%|█▋ | 678/10000 [01:06<11:18, 13.75it/s] Update steps: 7%|█▋ | 680/10000 [01:06<11:13, 13.84it/s] Update steps: 7%|█▋ | 682/10000 [01:06<11:10, 13.90it/s] Update steps: 7%|█▋ | 684/10000 [01:06<11:12, 13.84it/s] Update steps: 7%|█▋ | 686/10000 [01:06<11:17, 13.74it/s] Update steps: 7%|█▋ | 688/10000 [01:06<11:22, 13.65it/s] Update steps: 7%|█▋ | 690/10000 [01:06<11:21, 13.67it/s] Update steps: 7%|█▋ | 692/10000 [01:07<11:17, 13.74it/s] Update steps: 7%|█▋ | 694/10000 [01:07<11:13, 13.83it/s] Update steps: 7%|█▋ | 696/10000 [01:07<11:10, 13.87it/s] Update steps: 7%|█▋ | 698/10000 [01:07<11:06, 13.96it/s] Update steps: 7%|█▊ | 700/10000 [01:07<11:15, 13.77it/s] Update steps: 7%|█▊ | 702/10000 [01:07<11:20, 13.67it/s] Update steps: 7%|█▊ | 704/10000 [01:07<11:23, 13.61it/s] Update steps: 7%|█▊ | 706/10000 [01:08<11:22, 13.61it/s] Update steps: 7%|█▊ | 708/10000 [01:08<11:19, 13.68it/s] Update steps: 7%|█▊ | 710/10000 [01:08<11:15, 13.75it/s] Update steps: 7%|█▊ | 712/10000 [01:08<11:13, 13.79it/s] Update steps: 7%|█▊ | 714/10000 [01:08<11:15, 13.76it/s] Update steps: 7%|█▊ | 716/10000 [01:08<11:11, 13.84it/s] Update steps: 7%|█▊ | 718/10000 [01:08<11:06, 13.93it/s] Update steps: 7%|█▊ | 720/10000 [01:09<11:04, 13.97it/s] Update steps: 7%|█▊ | 722/10000 [01:09<11:07, 13.91it/s] Update steps: 7%|█▊ | 724/10000 [01:09<11:08, 13.88it/s] Update steps: 7%|█▊ | 726/10000 [01:09<11:11, 13.81it/s] Update steps: 7%|█▊ | 728/10000 [01:09<11:09, 13.86it/s] Update steps: 7%|█▊ | 730/10000 [01:09<11:06, 13.90it/s] Update steps: 7%|█▊ | 732/10000 [01:09<11:07, 13.89it/s] Update steps: 7%|█▊ | 734/10000 [01:10<11:06, 13.90it/s] Update steps: 7%|█▊ | 736/10000 [01:10<11:05, 13.93it/s] Update steps: 7%|█▊ | 738/10000 [01:10<11:01, 13.99it/s] Update steps: 7%|█▊ | 740/10000 [01:10<11:01, 14.00it/s] Update steps: 7%|█▊ | 742/10000 [01:10<11:07, 13.87it/s] Update steps: 7%|█▊ | 744/10000 [01:10<11:09, 13.83it/s] Update steps: 7%|█▊ | 746/10000 [01:10<11:17, 13.67it/s] Update steps: 7%|█▊ | 748/10000 [01:11<11:14, 13.71it/s] Update steps: 8%|█▉ | 750/10000 [01:11<11:15, 13.70it/s] Update steps: 8%|█▉ | 752/10000 [01:11<11:16, 13.67it/s] Update steps: 8%|█▉ | 754/10000 [01:11<11:11, 13.77it/s] Update steps: 8%|█▉ | 756/10000 [01:11<11:08, 13.83it/s] Update steps: 8%|█▉ | 758/10000 [01:11<11:10, 13.79it/s] Update steps: 8%|█▉ | 760/10000 [01:11<11:11, 13.75it/s] Update steps: 8%|█▉ | 762/10000 [01:12<11:08, 13.81it/s] Update steps: 8%|█▉ | 764/10000 [01:12<11:12, 13.73it/s] Update steps: 8%|█▉ | 766/10000 [01:12<11:09, 13.79it/s] Update steps: 8%|█▉ | 768/10000 [01:12<11:09, 13.79it/s] Update steps: 8%|█▉ | 770/10000 [01:12<11:05, 13.88it/s] Update steps: 8%|█▉ | 772/10000 [01:12<11:06, 13.85it/s] Update steps: 8%|█▉ | 774/10000 [01:12<11:08, 13.81it/s] Update steps: 8%|█▉ | 776/10000 [01:13<11:08, 13.79it/s] Update steps: 8%|█▉ | 778/10000 [01:13<11:13, 13.70it/s] Update steps: 8%|█▉ | 780/10000 [01:13<11:21, 13.53it/s] Update steps: 8%|█▉ | 782/10000 [01:13<11:23, 13.49it/s] Update steps: 8%|█▉ | 784/10000 [01:13<11:16, 13.63it/s] Update steps: 8%|█▉ | 786/10000 [01:13<11:16, 13.62it/s] Update steps: 8%|█▉ | 788/10000 [01:14<11:25, 13.43it/s] Update steps: 8%|█▉ | 790/10000 [01:14<11:25, 13.43it/s] Update steps: 8%|█▉ | 792/10000 [01:14<11:17, 13.59it/s] Update steps: 8%|█▉ | 794/10000 [01:14<11:14, 13.66it/s] Update steps: 8%|█▉ | 796/10000 [01:14<11:13, 13.66it/s] Update steps: 8%|█▉ | 798/10000 [01:14<11:18, 13.57it/s] Update steps: 8%|██ | 800/10000 [01:14<11:08, 13.75it/s] Update steps: 8%|██ | 802/10000 [01:15<11:06, 13.80it/s] Update steps: 8%|██ | 804/10000 [01:15<11:02, 13.88it/s] Update steps: 8%|██ | 806/10000 [01:15<11:06, 13.80it/s] Update steps: 8%|██ | 808/10000 [01:15<11:03, 13.85it/s] Update steps: 8%|██ | 810/10000 [01:15<11:05, 13.81it/s] Update steps: 8%|██ | 812/10000 [01:15<11:03, 13.85it/s] Update steps: 8%|██ | 814/10000 [01:15<11:03, 13.84it/s] Update steps: 8%|██ | 816/10000 [01:16<11:11, 13.68it/s] Update steps: 8%|██ | 818/10000 [01:16<11:05, 13.79it/s] Update steps: 8%|██ | 820/10000 [01:16<11:12, 13.66it/s] Update steps: 8%|██ | 822/10000 [01:16<11:05, 13.78it/s] Update steps: 8%|██ | 824/10000 [01:16<11:00, 13.89it/s] Update steps: 8%|██ | 826/10000 [01:16<10:58, 13.93it/s] Update steps: 8%|██ | 828/10000 [01:16<10:57, 13.95it/s] Update steps: 8%|██ | 830/10000 [01:17<10:58, 13.92it/s] Update steps: 8%|██ | 832/10000 [01:17<10:59, 13.91it/s] Update steps: 8%|██ | 834/10000 [01:17<10:59, 13.91it/s] Update steps: 8%|██ | 836/10000 [01:17<10:59, 13.90it/s] Update steps: 8%|██ | 838/10000 [01:17<10:55, 13.97it/s] Update steps: 8%|██ | 840/10000 [01:17<10:55, 13.97it/s] Update steps: 8%|██ | 842/10000 [01:17<10:53, 14.02it/s] Update steps: 8%|██ | 844/10000 [01:18<10:51, 14.05it/s] Update steps: 8%|██ | 846/10000 [01:18<10:54, 14.00it/s] Update steps: 8%|██ | 848/10000 [01:18<10:56, 13.94it/s] Update steps: 8%|██▏ | 850/10000 [01:18<11:00, 13.85it/s] Update steps: 9%|██▏ | 852/10000 [01:18<11:01, 13.84it/s] Update steps: 9%|██▏ | 854/10000 [01:18<11:02, 13.80it/s] Update steps: 9%|██▏ | 856/10000 [01:18<11:06, 13.72it/s] Update steps: 9%|██▏ | 858/10000 [01:19<11:10, 13.63it/s] Update steps: 9%|██▏ | 860/10000 [01:19<11:08, 13.67it/s] Update steps: 9%|██▏ | 862/10000 [01:19<11:08, 13.67it/s] Update steps: 9%|██▏ | 864/10000 [01:19<11:08, 13.66it/s] Update steps: 9%|██▏ | 866/10000 [01:19<11:07, 13.68it/s] Update steps: 9%|██▏ | 868/10000 [01:19<11:03, 13.77it/s] Update steps: 9%|██▏ | 870/10000 [01:19<10:57, 13.88it/s] Update steps: 9%|██▏ | 872/10000 [01:20<10:57, 13.89it/s] Update steps: 9%|██▏ | 874/10000 [01:20<10:58, 13.85it/s] Update steps: 9%|██▏ | 876/10000 [01:20<11:02, 13.77it/s] Update steps: 9%|██▏ | 878/10000 [01:20<11:01, 13.80it/s] Update steps: 9%|██▏ | 880/10000 [01:20<10:59, 13.82it/s] Update steps: 9%|██▏ | 882/10000 [01:20<11:05, 13.70it/s] Update steps: 9%|██▏ | 884/10000 [01:20<11:05, 13.70it/s] Update steps: 9%|██▏ | 886/10000 [01:21<11:05, 13.69it/s] Update steps: 9%|██▏ | 888/10000 [01:21<11:03, 13.73it/s] Update steps: 9%|██▏ | 890/10000 [01:21<11:02, 13.74it/s] Update steps: 9%|██▏ | 892/10000 [01:21<10:59, 13.82it/s] Update steps: 9%|██▏ | 894/10000 [01:21<10:59, 13.80it/s] Update steps: 9%|██▏ | 896/10000 [01:21<11:05, 13.68it/s] Update steps: 9%|██▏ | 898/10000 [01:21<11:05, 13.68it/s] Update steps: 9%|██▎ | 900/10000 [01:22<11:01, 13.76it/s] Update steps: 9%|██▎ | 902/10000 [01:22<11:01, 13.76it/s] Update steps: 9%|██▎ | 904/10000 [01:22<10:58, 13.81it/s] Update steps: 9%|██▎ | 906/10000 [01:22<10:58, 13.81it/s] Update steps: 9%|██▎ | 908/10000 [01:22<11:05, 13.66it/s] Update steps: 9%|██▎ | 910/10000 [01:22<11:00, 13.76it/s] Update steps: 9%|██▎ | 912/10000 [01:23<10:59, 13.78it/s] Update steps: 9%|██▎ | 914/10000 [01:23<10:53, 13.90it/s] Update steps: 9%|██▎ | 916/10000 [01:23<10:55, 13.86it/s] Update steps: 9%|██▎ | 918/10000 [01:23<10:56, 13.83it/s] Update steps: 9%|██▎ | 920/10000 [01:23<10:56, 13.84it/s] Update steps: 9%|██▎ | 922/10000 [01:23<10:52, 13.91it/s] Update steps: 9%|██▎ | 924/10000 [01:23<10:50, 13.95it/s] Update steps: 9%|██▎ | 926/10000 [01:24<11:03, 13.67it/s] Update steps: 9%|██▎ | 928/10000 [01:24<11:05, 13.63it/s] Update steps: 9%|██▎ | 930/10000 [01:24<11:07, 13.59it/s] Update steps: 9%|██▎ | 932/10000 [01:24<11:05, 13.63it/s] Update steps: 9%|██▎ | 934/10000 [01:24<11:06, 13.61it/s] Update steps: 9%|██▎ | 936/10000 [01:24<11:05, 13.62it/s] Update steps: 9%|██▎ | 938/10000 [01:24<11:02, 13.68it/s] Update steps: 9%|██▎ | 940/10000 [01:25<10:57, 13.78it/s] Update steps: 9%|██▎ | 942/10000 [01:25<10:51, 13.90it/s] Update steps: 9%|██▎ | 944/10000 [01:25<10:48, 13.97it/s] Update steps: 9%|██▎ | 946/10000 [01:25<10:49, 13.94it/s] Update steps: 9%|██▎ | 948/10000 [01:25<10:49, 13.94it/s] Update steps: 10%|██▍ | 950/10000 [01:25<10:51, 13.88it/s] Update steps: 10%|██▍ | 952/10000 [01:25<10:57, 13.76it/s] Update steps: 10%|██▍ | 954/10000 [01:26<10:57, 13.76it/s] Update steps: 10%|██▍ | 956/10000 [01:26<10:52, 13.85it/s] Update steps: 10%|██▍ | 958/10000 [01:26<10:54, 13.81it/s] Update steps: 10%|██▍ | 960/10000 [01:26<10:57, 13.75it/s] Update steps: 10%|██▍ | 962/10000 [01:26<10:52, 13.86it/s] Update steps: 10%|██▍ | 964/10000 [01:26<10:59, 13.70it/s] Update steps: 10%|██▍ | 966/10000 [01:26<10:59, 13.69it/s] Update steps: 10%|██▍ | 968/10000 [01:27<10:59, 13.69it/s] Update steps: 10%|██▍ | 970/10000 [01:27<10:57, 13.74it/s] Update steps: 10%|██▍ | 972/10000 [01:27<10:52, 13.83it/s] Update steps: 10%|██▍ | 974/10000 [01:27<10:52, 13.83it/s] Update steps: 10%|██▍ | 976/10000 [01:27<10:50, 13.88it/s] Update steps: 10%|██▍ | 978/10000 [01:27<10:55, 13.76it/s] Update steps: 10%|██▍ | 980/10000 [01:27<10:53, 13.79it/s] Update steps: 10%|██▍ | 982/10000 [01:28<10:56, 13.73it/s] Update steps: 10%|██▍ | 984/10000 [01:28<10:51, 13.84it/s] Update steps: 10%|██▍ | 986/10000 [01:28<10:49, 13.87it/s] Update steps: 10%|██▍ | 988/10000 [01:28<10:45, 13.96it/s] Update steps: 10%|██▍ | 990/10000 [01:28<10:42, 14.02it/s] Update steps: 10%|██▍ | 992/10000 [01:28<10:43, 13.99it/s] Update steps: 10%|██▍ | 994/10000 [01:28<10:49, 13.86it/s] Update steps: 10%|██▍ | 996/10000 [01:29<10:50, 13.83it/s] Update steps: 10%|██▍ | 998/10000 [01:29<10:48, 13.89it/s] Update steps: 10%|██▍ | 1000/10000 [01:29<10:44, 13.95it/s]2026-04-17 10:54:49.763 | INFO | __main__:main:490 - Performing evaluation at step 1000 +2026-04-17 10:54:49.767 | INFO | utils.eval:evaluate_model:21 - Loading validation data from 8 local validation files +2026-04-17 10:54:50.518 | INFO | utils.eval:evaluate_model:43 - Loaded validation dataset in 0.75 seconds +2026-04-17 10:54:50.519 | INFO | utils.eval:evaluate_model:60 - Eval set prepared in 0.76 seconds + Update steps: 10%|██▍ | 1000/10000 [01:40<10:44, 13.95it/s]2026-04-17 10:55:30.032 | INFO | __main__:main:505 - Eval loss at step 1000: 4.161990797435933, Eval perplexity: 64.19920307365534 + Update steps: 10%|██ | 1001/10000 [02:09<17:56:37, 7.18s/it] Update steps: 10%|██ | 1003/10000 [02:09<11:59:07, 4.80s/it] Update steps: 10%|██▏ | 1005/10000 [02:10<8:09:40, 3.27s/it] Update steps: 10%|██▏ | 1007/10000 [02:10<5:38:06, 2.26s/it] Update steps: 10%|██▏ | 1009/10000 [02:10<3:56:09, 1.58s/it] Update steps: 10%|██▏ | 1011/10000 [02:10<2:46:47, 1.11s/it] Update steps: 10%|██▏ | 1013/10000 [02:10<1:59:06, 1.26it/s] Update steps: 10%|██▏ | 1015/10000 [02:10<1:26:14, 1.74it/s] Update steps: 10%|██▏ | 1017/10000 [02:10<1:03:26, 2.36it/s] Update steps: 10%|██▍ | 1019/10000 [02:11<47:33, 3.15it/s] Update steps: 10%|██▍ | 1021/10000 [02:11<36:35, 4.09it/s] Update steps: 10%|██▍ | 1023/10000 [02:11<28:45, 5.20it/s] Update steps: 10%|██▍ | 1025/10000 [02:11<23:20, 6.41it/s] Update steps: 10%|██▍ | 1027/10000 [02:11<19:34, 7.64it/s] Update steps: 10%|██▍ | 1029/10000 [02:11<16:58, 8.81it/s] Update steps: 10%|██▍ | 1031/10000 [02:11<15:08, 9.88it/s] Update steps: 10%|██▍ | 1033/10000 [02:12<13:49, 10.81it/s] Update steps: 10%|██▍ | 1035/10000 [02:12<13:03, 11.45it/s] Update steps: 10%|██▍ | 1037/10000 [02:12<12:24, 12.04it/s] Update steps: 10%|██▍ | 1039/10000 [02:12<11:54, 12.54it/s] Update steps: 10%|██▍ | 1041/10000 [02:12<11:30, 12.97it/s] Update steps: 10%|██▌ | 1043/10000 [02:12<11:26, 13.05it/s] Update steps: 10%|██▌ | 1045/10000 [02:12<11:16, 13.24it/s] Update steps: 10%|██▌ | 1047/10000 [02:13<11:02, 13.50it/s] Update steps: 10%|██▌ | 1049/10000 [02:13<10:53, 13.69it/s] Update steps: 11%|██▌ | 1051/10000 [02:13<10:53, 13.70it/s] Update steps: 11%|██▌ | 1053/10000 [02:13<10:48, 13.79it/s] Update steps: 11%|██▌ | 1055/10000 [02:13<10:46, 13.83it/s] Update steps: 11%|██▌ | 1057/10000 [02:13<10:49, 13.77it/s] Update steps: 11%|██▌ | 1059/10000 [02:13<10:49, 13.76it/s] Update steps: 11%|██▌ | 1061/10000 [02:14<10:51, 13.73it/s] Update steps: 11%|██▌ | 1063/10000 [02:14<10:44, 13.86it/s] Update steps: 11%|██▌ | 1065/10000 [02:14<10:46, 13.82it/s] Update steps: 11%|██▌ | 1067/10000 [02:14<10:43, 13.87it/s] Update steps: 11%|██▌ | 1069/10000 [02:14<10:47, 13.80it/s] Update steps: 11%|██▌ | 1071/10000 [02:14<10:44, 13.85it/s] Update steps: 11%|██▌ | 1073/10000 [02:14<10:53, 13.67it/s] Update steps: 11%|██▌ | 1075/10000 [02:15<10:51, 13.70it/s] Update steps: 11%|██▌ | 1077/10000 [02:15<10:46, 13.79it/s] Update steps: 11%|██▌ | 1079/10000 [02:15<10:42, 13.89it/s] Update steps: 11%|██▌ | 1081/10000 [02:15<10:45, 13.82it/s] Update steps: 11%|██▌ | 1083/10000 [02:15<10:47, 13.78it/s] Update steps: 11%|██▌ | 1085/10000 [02:15<10:47, 13.77it/s] Update steps: 11%|██▌ | 1087/10000 [02:15<10:42, 13.87it/s] Update steps: 11%|██▌ | 1089/10000 [02:16<10:39, 13.93it/s] Update steps: 11%|██▌ | 1091/10000 [02:16<10:36, 14.00it/s] Update steps: 11%|██▌ | 1093/10000 [02:16<10:47, 13.76it/s] Update steps: 11%|██▋ | 1095/10000 [02:16<10:43, 13.84it/s] Update steps: 11%|██▋ | 1097/10000 [02:16<10:39, 13.92it/s] Update steps: 11%|██▋ | 1099/10000 [02:16<10:37, 13.97it/s] Update steps: 11%|██▋ | 1101/10000 [02:16<10:39, 13.92it/s] Update steps: 11%|██▋ | 1103/10000 [02:17<10:39, 13.92it/s] Update steps: 11%|██▋ | 1105/10000 [02:17<10:39, 13.91it/s] Update steps: 11%|██▋ | 1107/10000 [02:17<10:37, 13.96it/s] Update steps: 11%|██▋ | 1109/10000 [02:17<10:36, 13.98it/s] Update steps: 11%|██▋ | 1111/10000 [02:17<10:33, 14.03it/s] Update steps: 11%|██▋ | 1113/10000 [02:17<10:41, 13.85it/s] Update steps: 11%|██▋ | 1115/10000 [02:17<10:45, 13.76it/s] Update steps: 11%|██▋ | 1117/10000 [02:18<10:38, 13.90it/s] Update steps: 11%|██▋ | 1119/10000 [02:18<10:39, 13.90it/s] Update steps: 11%|██▋ | 1121/10000 [02:18<10:38, 13.90it/s] Update steps: 11%|██▋ | 1123/10000 [02:18<10:37, 13.92it/s] Update steps: 11%|██▋ | 1125/10000 [02:18<10:38, 13.89it/s] Update steps: 11%|██▋ | 1127/10000 [02:18<10:35, 13.97it/s] Update steps: 11%|██▋ | 1129/10000 [02:18<10:32, 14.02it/s] Update steps: 11%|██▋ | 1131/10000 [02:19<10:32, 14.02it/s] Update steps: 11%|██▋ | 1133/10000 [02:19<10:35, 13.96it/s] Update steps: 11%|██▋ | 1135/10000 [02:19<10:35, 13.94it/s] Update steps: 11%|██▋ | 1137/10000 [02:19<10:35, 13.94it/s] Update steps: 11%|██▋ | 1139/10000 [02:19<10:38, 13.87it/s] Update steps: 11%|██▋ | 1141/10000 [02:19<10:36, 13.91it/s] Update steps: 11%|██▋ | 1143/10000 [02:19<10:38, 13.87it/s] Update steps: 11%|██▋ | 1145/10000 [02:20<10:42, 13.78it/s] Update steps: 11%|██▊ | 1147/10000 [02:20<10:43, 13.77it/s] Update steps: 11%|██▊ | 1149/10000 [02:20<10:38, 13.85it/s] Update steps: 12%|██▊ | 1151/10000 [02:20<10:41, 13.79it/s] Update steps: 12%|██▊ | 1153/10000 [02:20<10:41, 13.80it/s] Update steps: 12%|██▊ | 1155/10000 [02:20<10:36, 13.89it/s] Update steps: 12%|██▊ | 1157/10000 [02:20<10:34, 13.93it/s] Update steps: 12%|██▊ | 1159/10000 [02:21<10:39, 13.82it/s] Update steps: 12%|██▊ | 1161/10000 [02:21<10:43, 13.74it/s] Update steps: 12%|██▊ | 1163/10000 [02:21<10:40, 13.79it/s] Update steps: 12%|██▊ | 1165/10000 [02:21<10:37, 13.85it/s] Update steps: 12%|██▊ | 1167/10000 [02:21<10:40, 13.79it/s] Update steps: 12%|██▊ | 1169/10000 [02:21<10:36, 13.87it/s] Update steps: 12%|██▊ | 1171/10000 [02:21<10:34, 13.91it/s] Update steps: 12%|██▊ | 1173/10000 [02:22<10:31, 13.97it/s] Update steps: 12%|██▊ | 1175/10000 [02:22<10:39, 13.80it/s] Update steps: 12%|██▊ | 1177/10000 [02:22<10:35, 13.88it/s] Update steps: 12%|██▊ | 1179/10000 [02:22<10:39, 13.79it/s] Update steps: 12%|██▊ | 1181/10000 [02:22<10:37, 13.84it/s] Update steps: 12%|██▊ | 1183/10000 [02:22<10:34, 13.89it/s] Update steps: 12%|██▊ | 1185/10000 [02:22<10:37, 13.82it/s] Update steps: 12%|██▊ | 1187/10000 [02:23<10:34, 13.88it/s] Update steps: 12%|██▊ | 1189/10000 [02:23<10:39, 13.77it/s] Update steps: 12%|██▊ | 1191/10000 [02:23<10:38, 13.80it/s] Update steps: 12%|██▊ | 1193/10000 [02:23<10:45, 13.65it/s] Update steps: 12%|██▊ | 1195/10000 [02:23<10:42, 13.70it/s] Update steps: 12%|██▊ | 1197/10000 [02:23<10:36, 13.83it/s] Update steps: 12%|██▉ | 1199/10000 [02:24<10:39, 13.77it/s] Update steps: 12%|██▉ | 1201/10000 [02:24<10:42, 13.69it/s] Update steps: 12%|██▉ | 1203/10000 [02:24<10:36, 13.82it/s] Update steps: 12%|██▉ | 1205/10000 [02:24<10:35, 13.84it/s] Update steps: 12%|██▉ | 1207/10000 [02:24<10:38, 13.77it/s] Update steps: 12%|██▉ | 1209/10000 [02:24<10:38, 13.77it/s] Update steps: 12%|██▉ | 1211/10000 [02:24<10:40, 13.73it/s] Update steps: 12%|██▉ | 1213/10000 [02:25<10:43, 13.67it/s] Update steps: 12%|██▉ | 1215/10000 [02:25<10:47, 13.56it/s] Update steps: 12%|██▉ | 1217/10000 [02:25<10:54, 13.42it/s] Update steps: 12%|██▉ | 1219/10000 [02:25<10:51, 13.48it/s] Update steps: 12%|██▉ | 1221/10000 [02:25<10:43, 13.65it/s] Update steps: 12%|██▉ | 1223/10000 [02:25<10:42, 13.65it/s] Update steps: 12%|██▉ | 1225/10000 [02:25<10:40, 13.70it/s] Update steps: 12%|██▉ | 1227/10000 [02:26<10:33, 13.85it/s] Update steps: 12%|██▉ | 1229/10000 [02:26<10:33, 13.85it/s] Update steps: 12%|██▉ | 1231/10000 [02:26<10:41, 13.67it/s] Update steps: 12%|██▉ | 1233/10000 [02:26<10:45, 13.58it/s] Update steps: 12%|██▉ | 1235/10000 [02:26<10:47, 13.54it/s] Update steps: 12%|██▉ | 1237/10000 [02:26<10:41, 13.66it/s] Update steps: 12%|██▉ | 1239/10000 [02:26<10:42, 13.63it/s] Update steps: 12%|██▉ | 1241/10000 [02:27<10:42, 13.63it/s] Update steps: 12%|██▉ | 1243/10000 [02:27<10:35, 13.78it/s] Update steps: 12%|██▉ | 1245/10000 [02:27<10:38, 13.71it/s] Update steps: 12%|██▉ | 1247/10000 [02:27<10:37, 13.74it/s] Update steps: 12%|██▉ | 1249/10000 [02:27<10:40, 13.67it/s] Update steps: 13%|███ | 1251/10000 [02:27<10:43, 13.59it/s] Update steps: 13%|███ | 1253/10000 [02:27<10:41, 13.65it/s] Update steps: 13%|███ | 1255/10000 [02:28<10:44, 13.57it/s] Update steps: 13%|███ | 1257/10000 [02:28<10:41, 13.63it/s] Update steps: 13%|███ | 1259/10000 [02:28<10:39, 13.66it/s] Update steps: 13%|███ | 1261/10000 [02:28<10:35, 13.76it/s] Update steps: 13%|███ | 1263/10000 [02:28<10:39, 13.67it/s] Update steps: 13%|███ | 1265/10000 [02:28<10:42, 13.60it/s] Update steps: 13%|███ | 1267/10000 [02:28<10:37, 13.69it/s] Update steps: 13%|███ | 1269/10000 [02:29<10:35, 13.74it/s] Update steps: 13%|███ | 1271/10000 [02:29<10:41, 13.60it/s] Update steps: 13%|███ | 1273/10000 [02:29<10:44, 13.54it/s] Update steps: 13%|███ | 1275/10000 [02:29<10:37, 13.69it/s] Update steps: 13%|███ | 1277/10000 [02:29<10:31, 13.82it/s] Update steps: 13%|███ | 1279/10000 [02:29<10:31, 13.82it/s] Update steps: 13%|███ | 1281/10000 [02:30<10:27, 13.90it/s] Update steps: 13%|███ | 1283/10000 [02:30<10:30, 13.82it/s] Update steps: 13%|███ | 1285/10000 [02:30<10:30, 13.82it/s] Update steps: 13%|███ | 1287/10000 [02:30<10:35, 13.72it/s] Update steps: 13%|███ | 1289/10000 [02:30<10:29, 13.83it/s] Update steps: 13%|███ | 1291/10000 [02:30<10:26, 13.90it/s] Update steps: 13%|███ | 1293/10000 [02:30<10:24, 13.94it/s] Update steps: 13%|███ | 1295/10000 [02:31<10:24, 13.93it/s] Update steps: 13%|███ | 1297/10000 [02:31<10:30, 13.80it/s] Update steps: 13%|███ | 1299/10000 [02:31<10:32, 13.75it/s] Update steps: 13%|███ | 1301/10000 [02:31<10:28, 13.83it/s] Update steps: 13%|███▏ | 1303/10000 [02:31<10:28, 13.85it/s] Update steps: 13%|███▏ | 1305/10000 [02:31<10:30, 13.79it/s] Update steps: 13%|███▏ | 1307/10000 [02:31<10:32, 13.74it/s] Update steps: 13%|███▏ | 1309/10000 [02:32<10:27, 13.85it/s] Update steps: 13%|███▏ | 1311/10000 [02:32<10:28, 13.82it/s] Update steps: 13%|███▏ | 1313/10000 [02:32<10:32, 13.74it/s] Update steps: 13%|███▏ | 1315/10000 [02:32<10:28, 13.82it/s] Update steps: 13%|███▏ | 1317/10000 [02:32<10:28, 13.82it/s] Update steps: 13%|███▏ | 1319/10000 [02:32<10:25, 13.87it/s] Update steps: 13%|███▏ | 1321/10000 [02:32<10:28, 13.80it/s] Update steps: 13%|███▏ | 1323/10000 [02:33<10:29, 13.78it/s] Update steps: 13%|███▏ | 1325/10000 [02:33<10:30, 13.75it/s] Update steps: 13%|███▏ | 1327/10000 [02:33<10:27, 13.82it/s] Update steps: 13%|███▏ | 1329/10000 [02:33<10:26, 13.83it/s] Update steps: 13%|███▏ | 1331/10000 [02:33<10:26, 13.84it/s] Update steps: 13%|███▏ | 1333/10000 [02:33<10:29, 13.77it/s] Update steps: 13%|███▏ | 1335/10000 [02:33<10:30, 13.75it/s] Update steps: 13%|███▏ | 1337/10000 [02:34<10:26, 13.84it/s] Update steps: 13%|███▏ | 1339/10000 [02:34<10:21, 13.93it/s] Update steps: 13%|███▏ | 1341/10000 [02:34<10:20, 13.95it/s] Update steps: 13%|███▏ | 1343/10000 [02:34<10:18, 13.99it/s] Update steps: 13%|███▏ | 1345/10000 [02:34<10:23, 13.88it/s] Update steps: 13%|███▏ | 1347/10000 [02:34<10:22, 13.90it/s] Update steps: 13%|███▏ | 1349/10000 [02:34<10:22, 13.89it/s] Update steps: 14%|███▏ | 1351/10000 [02:35<10:24, 13.85it/s] Update steps: 14%|███▏ | 1353/10000 [02:35<10:22, 13.88it/s] Update steps: 14%|███▎ | 1355/10000 [02:35<10:25, 13.81it/s] Update steps: 14%|███▎ | 1357/10000 [02:35<10:22, 13.90it/s] Update steps: 14%|███▎ | 1359/10000 [02:35<10:20, 13.92it/s] Update steps: 14%|███▎ | 1361/10000 [02:35<10:19, 13.94it/s] Update steps: 14%|███▎ | 1363/10000 [02:35<10:19, 13.95it/s] Update steps: 14%|███▎ | 1365/10000 [02:36<10:22, 13.87it/s] Update steps: 14%|███▎ | 1367/10000 [02:36<10:27, 13.76it/s] Update steps: 14%|███▎ | 1369/10000 [02:36<10:22, 13.86it/s] Update steps: 14%|███▎ | 1371/10000 [02:36<10:20, 13.91it/s] Update steps: 14%|███▎ | 1373/10000 [02:36<10:19, 13.93it/s] Update steps: 14%|███▎ | 1375/10000 [02:36<10:22, 13.86it/s] Update steps: 14%|███▎ | 1377/10000 [02:36<10:24, 13.80it/s] Update steps: 14%|███▎ | 1379/10000 [02:37<10:20, 13.88it/s] Update steps: 14%|███▎ | 1381/10000 [02:37<10:18, 13.94it/s] Update steps: 14%|███▎ | 1383/10000 [02:37<10:20, 13.88it/s] Update steps: 14%|███▎ | 1385/10000 [02:37<10:24, 13.80it/s] Update steps: 14%|███▎ | 1387/10000 [02:37<10:20, 13.88it/s] Update steps: 14%|███▎ | 1389/10000 [02:37<10:19, 13.90it/s] Update steps: 14%|███▎ | 1391/10000 [02:37<10:23, 13.81it/s] Update steps: 14%|███▎ | 1393/10000 [02:38<11:02, 13.00it/s] Update steps: 14%|███▎ | 1395/10000 [02:38<10:55, 13.13it/s] Update steps: 14%|███▎ | 1397/10000 [02:38<10:46, 13.31it/s] Update steps: 14%|███▎ | 1399/10000 [02:38<10:40, 13.44it/s] Update steps: 14%|███▎ | 1401/10000 [02:38<10:33, 13.58it/s] Update steps: 14%|███▎ | 1403/10000 [02:38<10:25, 13.74it/s] Update steps: 14%|███▎ | 1405/10000 [02:38<10:27, 13.70it/s] Update steps: 14%|███▍ | 1407/10000 [02:39<10:28, 13.67it/s] Update steps: 14%|███▍ | 1409/10000 [02:39<10:29, 13.65it/s] Update steps: 14%|███▍ | 1411/10000 [02:39<10:23, 13.77it/s] Update steps: 14%|███▍ | 1413/10000 [02:39<10:20, 13.85it/s] Update steps: 14%|███▍ | 1415/10000 [02:39<10:16, 13.93it/s] Update steps: 14%|███▍ | 1417/10000 [02:39<10:17, 13.90it/s] Update steps: 14%|███▍ | 1419/10000 [02:40<10:20, 13.83it/s] Update steps: 14%|███▍ | 1421/10000 [02:40<10:19, 13.85it/s] Update steps: 14%|███▍ | 1423/10000 [02:40<10:16, 13.90it/s] Update steps: 14%|███▍ | 1425/10000 [02:40<10:18, 13.87it/s] Update steps: 14%|███▍ | 1427/10000 [02:40<10:16, 13.90it/s] Update steps: 14%|███▍ | 1429/10000 [02:40<10:25, 13.70it/s] Update steps: 14%|███▍ | 1431/10000 [02:40<10:21, 13.78it/s] Update steps: 14%|███▍ | 1433/10000 [02:41<10:26, 13.66it/s] Update steps: 14%|███▍ | 1435/10000 [02:41<10:22, 13.76it/s] Update steps: 14%|███▍ | 1437/10000 [02:41<10:17, 13.86it/s] Update steps: 14%|███▍ | 1439/10000 [02:41<10:14, 13.94it/s] Update steps: 14%|███▍ | 1441/10000 [02:41<10:13, 13.96it/s] Update steps: 14%|███▍ | 1443/10000 [02:41<10:16, 13.87it/s] Update steps: 14%|███▍ | 1445/10000 [02:41<10:24, 13.70it/s] Update steps: 14%|███▍ | 1447/10000 [02:42<10:26, 13.66it/s] Update steps: 14%|███▍ | 1449/10000 [02:42<10:25, 13.67it/s] Update steps: 15%|███▍ | 1451/10000 [02:42<10:23, 13.72it/s] Update steps: 15%|███▍ | 1453/10000 [02:42<10:20, 13.77it/s] Update steps: 15%|███▍ | 1455/10000 [02:42<10:17, 13.84it/s] Update steps: 15%|███▍ | 1457/10000 [02:42<10:20, 13.77it/s] Update steps: 15%|███▌ | 1459/10000 [02:42<10:16, 13.85it/s] Update steps: 15%|███▌ | 1461/10000 [02:43<10:14, 13.90it/s] Update steps: 15%|███▌ | 1463/10000 [02:43<10:13, 13.90it/s] Update steps: 15%|███▌ | 1465/10000 [02:43<10:18, 13.79it/s] Update steps: 15%|███▌ | 1467/10000 [02:43<10:21, 13.73it/s] Update steps: 15%|███▌ | 1469/10000 [02:43<10:16, 13.84it/s] Update steps: 15%|███▌ | 1471/10000 [02:43<10:15, 13.86it/s] Update steps: 15%|███▌ | 1473/10000 [02:43<10:14, 13.89it/s] Update steps: 15%|███▌ | 1475/10000 [02:44<10:11, 13.94it/s] Update steps: 15%|███▌ | 1477/10000 [02:44<10:16, 13.82it/s] Update steps: 15%|███▌ | 1479/10000 [02:44<10:17, 13.81it/s] Update steps: 15%|███▌ | 1481/10000 [02:44<10:14, 13.87it/s] Update steps: 15%|███▌ | 1483/10000 [02:44<10:12, 13.90it/s] Update steps: 15%|███▌ | 1485/10000 [02:44<10:18, 13.77it/s] Update steps: 15%|███▌ | 1487/10000 [02:44<10:18, 13.75it/s] Update steps: 15%|███▌ | 1489/10000 [02:45<10:23, 13.64it/s] Update steps: 15%|███▌ | 1491/10000 [02:45<10:24, 13.62it/s] Update steps: 15%|███▌ | 1493/10000 [02:45<10:18, 13.76it/s] Update steps: 15%|███▌ | 1495/10000 [02:45<10:19, 13.72it/s] Update steps: 15%|███▌ | 1497/10000 [02:45<10:20, 13.70it/s] Update steps: 15%|███▌ | 1499/10000 [02:45<10:25, 13.58it/s] Update steps: 15%|███▌ | 1501/10000 [02:45<10:19, 13.71it/s] Update steps: 15%|███▌ | 1503/10000 [02:46<10:27, 13.55it/s] Update steps: 15%|███▌ | 1505/10000 [02:46<10:29, 13.50it/s] Update steps: 15%|███▌ | 1507/10000 [02:46<10:22, 13.65it/s] Update steps: 15%|███▌ | 1509/10000 [02:46<10:18, 13.74it/s] Update steps: 15%|███▋ | 1511/10000 [02:46<10:17, 13.74it/s] Update steps: 15%|███▋ | 1513/10000 [02:46<10:25, 13.58it/s] Update steps: 15%|███▋ | 1515/10000 [02:46<10:21, 13.66it/s] Update steps: 15%|███▋ | 1517/10000 [02:47<10:21, 13.64it/s] Update steps: 15%|███▋ | 1519/10000 [02:47<10:19, 13.69it/s] Update steps: 15%|███▋ | 1521/10000 [02:47<10:16, 13.76it/s] Update steps: 15%|███▋ | 1523/10000 [02:47<10:18, 13.71it/s] Update steps: 15%|███▋ | 1525/10000 [02:47<10:21, 13.63it/s] Update steps: 15%|███▋ | 1527/10000 [02:47<10:15, 13.76it/s] Update steps: 15%|███▋ | 1529/10000 [02:48<10:17, 13.72it/s] Update steps: 15%|███▋ | 1531/10000 [02:48<10:17, 13.72it/s] Update steps: 15%|███▋ | 1533/10000 [02:48<10:16, 13.74it/s] Update steps: 15%|███▋ | 1535/10000 [02:48<10:15, 13.76it/s] Update steps: 15%|███▋ | 1537/10000 [02:48<10:19, 13.67it/s] Update steps: 15%|███▋ | 1539/10000 [02:48<10:18, 13.69it/s] Update steps: 15%|███▋ | 1541/10000 [02:48<10:18, 13.68it/s] Update steps: 15%|███▋ | 1543/10000 [02:49<10:18, 13.67it/s] Update steps: 15%|███▋ | 1545/10000 [02:49<10:16, 13.71it/s] Update steps: 15%|███▋ | 1547/10000 [02:49<10:11, 13.82it/s] Update steps: 15%|███▋ | 1549/10000 [02:49<10:09, 13.86it/s] Update steps: 16%|███▋ | 1551/10000 [02:49<10:12, 13.79it/s] Update steps: 16%|███▋ | 1553/10000 [02:49<10:12, 13.78it/s] Update steps: 16%|███▋ | 1555/10000 [02:49<10:16, 13.70it/s] Update steps: 16%|███▋ | 1557/10000 [02:50<10:18, 13.66it/s] Update steps: 16%|███▋ | 1559/10000 [02:50<10:17, 13.67it/s] Update steps: 16%|███▋ | 1561/10000 [02:50<10:15, 13.71it/s] Update steps: 16%|███▊ | 1563/10000 [02:50<10:11, 13.79it/s] Update steps: 16%|███▊ | 1565/10000 [02:50<10:11, 13.80it/s] Update steps: 16%|███▊ | 1567/10000 [02:50<10:08, 13.85it/s] Update steps: 16%|███▊ | 1569/10000 [02:50<10:04, 13.95it/s] Update steps: 16%|███▊ | 1571/10000 [02:51<10:09, 13.82it/s] Update steps: 16%|███▊ | 1573/10000 [02:51<10:10, 13.81it/s] Update steps: 16%|███▊ | 1575/10000 [02:51<10:10, 13.81it/s] Update steps: 16%|███▊ | 1577/10000 [02:51<10:05, 13.91it/s] Update steps: 16%|███▊ | 1579/10000 [02:51<10:02, 13.97it/s] Update steps: 16%|███▊ | 1581/10000 [02:51<10:12, 13.75it/s] Update steps: 16%|███▊ | 1583/10000 [02:51<10:10, 13.79it/s] Update steps: 16%|███▊ | 1585/10000 [02:52<10:12, 13.74it/s] Update steps: 16%|███▊ | 1587/10000 [02:52<10:07, 13.84it/s] Update steps: 16%|███▊ | 1589/10000 [02:52<10:07, 13.84it/s] Update steps: 16%|███▊ | 1591/10000 [02:52<10:05, 13.89it/s] Update steps: 16%|███▊ | 1593/10000 [02:52<10:13, 13.71it/s] Update steps: 16%|███▊ | 1595/10000 [02:52<10:13, 13.71it/s] Update steps: 16%|███▊ | 1597/10000 [02:52<10:14, 13.68it/s] Update steps: 16%|███▊ | 1599/10000 [02:53<10:11, 13.75it/s] Update steps: 16%|███▊ | 1601/10000 [02:53<10:16, 13.61it/s] Update steps: 16%|███▊ | 1603/10000 [02:53<10:13, 13.69it/s] Update steps: 16%|███▊ | 1605/10000 [02:53<10:12, 13.70it/s] Update steps: 16%|███▊ | 1607/10000 [02:53<10:09, 13.78it/s] Update steps: 16%|███▊ | 1609/10000 [02:53<10:10, 13.74it/s] Update steps: 16%|███▊ | 1611/10000 [02:53<10:09, 13.76it/s] Update steps: 16%|███▊ | 1613/10000 [02:54<10:11, 13.73it/s] Update steps: 16%|███▉ | 1615/10000 [02:54<10:07, 13.81it/s] Update steps: 16%|███▉ | 1617/10000 [02:54<10:07, 13.79it/s] Update steps: 16%|███▉ | 1619/10000 [02:54<10:07, 13.80it/s] Update steps: 16%|███▉ | 1621/10000 [02:54<10:02, 13.91it/s] Update steps: 16%|███▉ | 1623/10000 [02:54<10:09, 13.74it/s] Update steps: 16%|███▉ | 1625/10000 [02:54<10:04, 13.85it/s] Update steps: 16%|███▉ | 1627/10000 [02:55<10:02, 13.90it/s] Update steps: 16%|███▉ | 1629/10000 [02:55<10:04, 13.84it/s] Update steps: 16%|███▉ | 1631/10000 [02:55<10:04, 13.84it/s] Update steps: 16%|███▉ | 1633/10000 [02:55<10:03, 13.87it/s] Update steps: 16%|███▉ | 1635/10000 [02:55<10:06, 13.80it/s] Update steps: 16%|███▉ | 1637/10000 [02:55<10:01, 13.89it/s] Update steps: 16%|███▉ | 1639/10000 [02:55<10:00, 13.93it/s] Update steps: 16%|███▉ | 1641/10000 [02:56<09:58, 13.98it/s] Update steps: 16%|███▉ | 1643/10000 [02:56<10:03, 13.85it/s] Update steps: 16%|███▉ | 1645/10000 [02:56<10:06, 13.78it/s] Update steps: 16%|███▉ | 1647/10000 [02:56<10:05, 13.79it/s] Update steps: 16%|███▉ | 1649/10000 [02:56<10:01, 13.88it/s] Update steps: 17%|███▉ | 1651/10000 [02:56<10:00, 13.90it/s] Update steps: 17%|███▉ | 1653/10000 [02:56<09:58, 13.96it/s] Update steps: 17%|███▉ | 1655/10000 [02:57<09:56, 13.99it/s] Update steps: 17%|███▉ | 1657/10000 [02:57<09:55, 14.01it/s] Update steps: 17%|███▉ | 1659/10000 [02:57<09:55, 14.01it/s] Update steps: 17%|███▉ | 1661/10000 [02:57<10:01, 13.86it/s] Update steps: 17%|███▉ | 1663/10000 [02:57<10:00, 13.88it/s] Update steps: 17%|███▉ | 1665/10000 [02:57<10:04, 13.79it/s] Update steps: 17%|████ | 1667/10000 [02:58<10:00, 13.88it/s] Update steps: 17%|████ | 1669/10000 [02:58<10:06, 13.73it/s] Update steps: 17%|████ | 1671/10000 [02:58<10:11, 13.62it/s] Update steps: 17%|████ | 1673/10000 [02:58<10:16, 13.50it/s] Update steps: 17%|████ | 1675/10000 [02:58<10:19, 13.43it/s] Update steps: 17%|████ | 1677/10000 [02:58<10:18, 13.46it/s] Update steps: 17%|████ | 1679/10000 [02:58<10:10, 13.64it/s] Update steps: 17%|████ | 1681/10000 [02:59<10:07, 13.69it/s] Update steps: 17%|████ | 1683/10000 [02:59<10:11, 13.61it/s] Update steps: 17%|████ | 1685/10000 [02:59<10:06, 13.71it/s] Update steps: 17%|████ | 1687/10000 [02:59<10:10, 13.63it/s] Update steps: 17%|████ | 1689/10000 [02:59<10:09, 13.64it/s] Update steps: 17%|████ | 1691/10000 [02:59<10:06, 13.71it/s] Update steps: 17%|████ | 1693/10000 [02:59<10:01, 13.82it/s] Update steps: 17%|████ | 1695/10000 [03:00<10:07, 13.67it/s] Update steps: 17%|████ | 1697/10000 [03:00<10:03, 13.75it/s] Update steps: 17%|████ | 1699/10000 [03:00<10:08, 13.64it/s] Update steps: 17%|████ | 1701/10000 [03:00<10:02, 13.78it/s] Update steps: 17%|████ | 1703/10000 [03:00<09:58, 13.87it/s] Update steps: 17%|████ | 1705/10000 [03:00<10:04, 13.71it/s] Update steps: 17%|████ | 1707/10000 [03:00<10:00, 13.81it/s] Update steps: 17%|████ | 1709/10000 [03:01<10:03, 13.73it/s] Update steps: 17%|████ | 1711/10000 [03:01<10:00, 13.80it/s] Update steps: 17%|████ | 1713/10000 [03:01<10:03, 13.73it/s] Update steps: 17%|████ | 1715/10000 [03:01<10:07, 13.63it/s] Update steps: 17%|████ | 1717/10000 [03:01<10:06, 13.66it/s] Update steps: 17%|████▏ | 1719/10000 [03:01<10:13, 13.50it/s] Update steps: 17%|████▏ | 1721/10000 [03:01<10:09, 13.58it/s] Update steps: 17%|████▏ | 1723/10000 [03:02<10:05, 13.66it/s] Update steps: 17%|████▏ | 1725/10000 [03:02<10:04, 13.69it/s] Update steps: 17%|████▏ | 1727/10000 [03:02<10:03, 13.71it/s] Update steps: 17%|████▏ | 1729/10000 [03:02<10:04, 13.69it/s] Update steps: 17%|████▏ | 1731/10000 [03:02<10:01, 13.74it/s] Update steps: 17%|████▏ | 1733/10000 [03:02<09:58, 13.82it/s] Update steps: 17%|████▏ | 1735/10000 [03:02<09:56, 13.85it/s] Update steps: 17%|████▏ | 1737/10000 [03:03<09:54, 13.91it/s] Update steps: 17%|████▏ | 1739/10000 [03:03<09:52, 13.94it/s] Update steps: 17%|████▏ | 1741/10000 [03:03<09:51, 13.97it/s] Update steps: 17%|████▏ | 1743/10000 [03:03<09:50, 13.97it/s] Update steps: 17%|████▏ | 1745/10000 [03:03<09:58, 13.79it/s] Update steps: 17%|████▏ | 1747/10000 [03:03<10:00, 13.74it/s] Update steps: 17%|████▏ | 1749/10000 [03:03<10:02, 13.70it/s] Update steps: 18%|████▏ | 1751/10000 [03:04<10:01, 13.72it/s] Update steps: 18%|████▏ | 1753/10000 [03:04<10:06, 13.61it/s] Update steps: 18%|████▏ | 1755/10000 [03:04<10:01, 13.70it/s] Update steps: 18%|████▏ | 1757/10000 [03:04<10:05, 13.60it/s] Update steps: 18%|████▏ | 1759/10000 [03:04<10:06, 13.59it/s] Update steps: 18%|████▏ | 1761/10000 [03:04<10:10, 13.49it/s] Update steps: 18%|████▏ | 1763/10000 [03:05<10:04, 13.63it/s] Update steps: 18%|████▏ | 1765/10000 [03:05<10:00, 13.72it/s] Update steps: 18%|████▏ | 1767/10000 [03:05<10:02, 13.67it/s] Update steps: 18%|████▏ | 1769/10000 [03:05<10:04, 13.61it/s] Update steps: 18%|████▎ | 1771/10000 [03:05<10:03, 13.64it/s] Update steps: 18%|████▎ | 1773/10000 [03:05<10:03, 13.62it/s] Update steps: 18%|████▎ | 1775/10000 [03:05<10:01, 13.67it/s] Update steps: 18%|████▎ | 1777/10000 [03:06<09:56, 13.79it/s] Update steps: 18%|████▎ | 1779/10000 [03:06<09:52, 13.87it/s] Update steps: 18%|████▎ | 1781/10000 [03:06<09:50, 13.92it/s] Update steps: 18%|████▎ | 1783/10000 [03:06<09:49, 13.94it/s] Update steps: 18%|████▎ | 1785/10000 [03:06<09:55, 13.79it/s] Update steps: 18%|████▎ | 1787/10000 [03:06<09:55, 13.79it/s] Update steps: 18%|████▎ | 1789/10000 [03:06<09:58, 13.72it/s] Update steps: 18%|████▎ | 1791/10000 [03:07<09:56, 13.77it/s] Update steps: 18%|████▎ | 1793/10000 [03:07<09:58, 13.71it/s] Update steps: 18%|████▎ | 1795/10000 [03:07<09:54, 13.80it/s] Update steps: 18%|████▎ | 1797/10000 [03:07<09:51, 13.87it/s] Update steps: 18%|████▎ | 1799/10000 [03:07<09:50, 13.90it/s] Update steps: 18%|████▎ | 1801/10000 [03:07<09:57, 13.71it/s] Update steps: 18%|████▎ | 1803/10000 [03:07<10:02, 13.62it/s] Update steps: 18%|████▎ | 1805/10000 [03:08<10:11, 13.40it/s] Update steps: 18%|████▎ | 1807/10000 [03:08<10:08, 13.47it/s] Update steps: 18%|████▎ | 1809/10000 [03:08<10:00, 13.64it/s] Update steps: 18%|████▎ | 1811/10000 [03:08<09:57, 13.70it/s] Update steps: 18%|████▎ | 1813/10000 [03:08<09:56, 13.72it/s] Update steps: 18%|████▎ | 1815/10000 [03:08<09:56, 13.71it/s] Update steps: 18%|████▎ | 1817/10000 [03:08<10:00, 13.62it/s] Update steps: 18%|████▎ | 1819/10000 [03:09<10:00, 13.63it/s] Update steps: 18%|████▎ | 1821/10000 [03:09<09:56, 13.71it/s] Update steps: 18%|████▍ | 1823/10000 [03:09<09:54, 13.76it/s] Update steps: 18%|████▍ | 1825/10000 [03:09<09:57, 13.67it/s] Update steps: 18%|████▍ | 1827/10000 [03:09<10:00, 13.62it/s] Update steps: 18%|████▍ | 1829/10000 [03:09<09:57, 13.67it/s] Update steps: 18%|████▍ | 1831/10000 [03:09<09:57, 13.67it/s] Update steps: 18%|████▍ | 1833/10000 [03:10<09:55, 13.71it/s] Update steps: 18%|████▍ | 1835/10000 [03:10<09:59, 13.61it/s] Update steps: 18%|████▍ | 1837/10000 [03:10<09:55, 13.70it/s] Update steps: 18%|████▍ | 1839/10000 [03:10<09:55, 13.70it/s] Update steps: 18%|████▍ | 1841/10000 [03:10<09:54, 13.71it/s] Update steps: 18%|████▍ | 1843/10000 [03:10<09:56, 13.67it/s] Update steps: 18%|████▍ | 1845/10000 [03:10<09:57, 13.65it/s] Update steps: 18%|████▍ | 1847/10000 [03:11<09:59, 13.60it/s] Update steps: 18%|████▍ | 1849/10000 [03:11<09:59, 13.59it/s] Update steps: 19%|████▍ | 1851/10000 [03:11<10:04, 13.47it/s] Update steps: 19%|████▍ | 1853/10000 [03:11<09:57, 13.64it/s] Update steps: 19%|████▍ | 1855/10000 [03:11<09:53, 13.72it/s] Update steps: 19%|████▍ | 1857/10000 [03:11<09:52, 13.74it/s] Update steps: 19%|████▍ | 1859/10000 [03:12<09:51, 13.76it/s] Update steps: 19%|████▍ | 1861/10000 [03:12<09:52, 13.73it/s] Update steps: 19%|████▍ | 1863/10000 [03:12<09:52, 13.73it/s] Update steps: 19%|████▍ | 1865/10000 [03:12<09:47, 13.84it/s] Update steps: 19%|████▍ | 1867/10000 [03:12<09:49, 13.79it/s] Update steps: 19%|████▍ | 1869/10000 [03:12<09:51, 13.75it/s] Update steps: 19%|████▍ | 1871/10000 [03:12<09:53, 13.70it/s] Update steps: 19%|████▍ | 1873/10000 [03:13<09:47, 13.83it/s] Update steps: 19%|████▌ | 1875/10000 [03:13<09:44, 13.90it/s] Update steps: 19%|████▌ | 1877/10000 [03:13<09:44, 13.89it/s] Update steps: 19%|████▌ | 1879/10000 [03:13<09:50, 13.74it/s] Update steps: 19%|████▌ | 1881/10000 [03:13<09:56, 13.62it/s] Update steps: 19%|████▌ | 1883/10000 [03:13<09:55, 13.62it/s] Update steps: 19%|████▌ | 1885/10000 [03:13<09:57, 13.58it/s] Update steps: 19%|████▌ | 1887/10000 [03:14<09:51, 13.71it/s] Update steps: 19%|████▌ | 1889/10000 [03:14<09:57, 13.58it/s] Update steps: 19%|████▌ | 1891/10000 [03:14<09:54, 13.63it/s] Update steps: 19%|████▌ | 1893/10000 [03:14<09:50, 13.72it/s] Update steps: 19%|████▌ | 1895/10000 [03:14<09:46, 13.81it/s] Update steps: 19%|████▌ | 1897/10000 [03:14<09:52, 13.67it/s] Update steps: 19%|████▌ | 1899/10000 [03:14<09:47, 13.80it/s] Update steps: 19%|████▌ | 1901/10000 [03:15<09:48, 13.77it/s] Update steps: 19%|████▌ | 1903/10000 [03:15<09:44, 13.85it/s] Update steps: 19%|████▌ | 1905/10000 [03:15<09:45, 13.83it/s] Update steps: 19%|████▌ | 1907/10000 [03:15<09:45, 13.82it/s] Update steps: 19%|████▌ | 1909/10000 [03:15<09:41, 13.91it/s] Update steps: 19%|████▌ | 1911/10000 [03:15<09:39, 13.95it/s] Update steps: 19%|████▌ | 1913/10000 [03:15<09:38, 13.98it/s] Update steps: 19%|████▌ | 1915/10000 [03:16<09:39, 13.94it/s] Update steps: 19%|████▌ | 1917/10000 [03:16<09:38, 13.97it/s] Update steps: 19%|████▌ | 1919/10000 [03:16<09:37, 14.00it/s] Update steps: 19%|████▌ | 1921/10000 [03:16<09:39, 13.93it/s] Update steps: 19%|████▌ | 1923/10000 [03:16<09:57, 13.52it/s] Update steps: 19%|████▌ | 1925/10000 [03:16<09:56, 13.55it/s] Update steps: 19%|████▌ | 1927/10000 [03:16<09:50, 13.67it/s] Update steps: 19%|████▋ | 1929/10000 [03:17<10:01, 13.42it/s] Update steps: 19%|████▋ | 1931/10000 [03:17<09:56, 13.54it/s] Update steps: 19%|████▋ | 1933/10000 [03:17<09:55, 13.55it/s] Update steps: 19%|████▋ | 1935/10000 [03:17<09:52, 13.62it/s] Update steps: 19%|████▋ | 1937/10000 [03:17<09:51, 13.64it/s] Update steps: 19%|████▋ | 1939/10000 [03:17<09:45, 13.77it/s] Update steps: 19%|████▋ | 1941/10000 [03:17<09:42, 13.83it/s] Update steps: 19%|████▋ | 1943/10000 [03:18<09:40, 13.88it/s] Update steps: 19%|████▋ | 1945/10000 [03:18<09:37, 13.94it/s] Update steps: 19%|████▋ | 1947/10000 [03:18<09:39, 13.90it/s] Update steps: 19%|████▋ | 1949/10000 [03:18<09:38, 13.93it/s] Update steps: 20%|████▋ | 1951/10000 [03:18<09:38, 13.91it/s] Update steps: 20%|████▋ | 1953/10000 [03:18<09:35, 13.98it/s] Update steps: 20%|████▋ | 1955/10000 [03:18<09:39, 13.89it/s] Update steps: 20%|████▋ | 1957/10000 [03:19<09:41, 13.83it/s] Update steps: 20%|████▋ | 1959/10000 [03:19<09:42, 13.79it/s] Update steps: 20%|████▋ | 1961/10000 [03:19<09:41, 13.82it/s] Update steps: 20%|████▋ | 1963/10000 [03:19<09:38, 13.90it/s] Update steps: 20%|████▋ | 1965/10000 [03:19<09:38, 13.89it/s] Update steps: 20%|████▋ | 1967/10000 [03:19<09:35, 13.97it/s] Update steps: 20%|████▋ | 1969/10000 [03:19<09:38, 13.87it/s] Update steps: 20%|████▋ | 1971/10000 [03:20<09:39, 13.86it/s] Update steps: 20%|████▋ | 1973/10000 [03:20<09:46, 13.70it/s] Update steps: 20%|████▋ | 1975/10000 [03:20<09:48, 13.64it/s] Update steps: 20%|████▋ | 1977/10000 [03:20<09:49, 13.61it/s] Update steps: 20%|████▋ | 1979/10000 [03:20<09:51, 13.55it/s] Update steps: 20%|████▊ | 1981/10000 [03:20<09:49, 13.60it/s] Update steps: 20%|████▊ | 1983/10000 [03:21<09:46, 13.67it/s] Update steps: 20%|████▊ | 1985/10000 [03:21<09:46, 13.67it/s] Update steps: 20%|████▊ | 1987/10000 [03:21<09:44, 13.71it/s] Update steps: 20%|████▊ | 1989/10000 [03:21<09:40, 13.80it/s] Update steps: 20%|████▊ | 1991/10000 [03:21<09:37, 13.86it/s] Update steps: 20%|████▊ | 1993/10000 [03:21<09:39, 13.82it/s] Update steps: 20%|████▊ | 1995/10000 [03:21<09:39, 13.81it/s] Update steps: 20%|████▊ | 1997/10000 [03:22<09:44, 13.70it/s] Update steps: 20%|████▊ | 1999/10000 [03:22<09:48, 13.60it/s]2026-04-17 10:56:42.656 | INFO | __main__:main:490 - Performing evaluation at step 2000 +2026-04-17 10:56:42.658 | INFO | utils.eval:evaluate_model:21 - Loading validation data from 8 local validation files +2026-04-17 10:56:43.393 | INFO | utils.eval:evaluate_model:43 - Loaded validation dataset in 0.74 seconds +2026-04-17 10:56:43.394 | INFO | utils.eval:evaluate_model:60 - Eval set prepared in 0.74 seconds + Update steps: 20%|████▊ | 2000/10000 [03:36<09:48, 13.60it/s]2026-04-17 10:57:18.661 | INFO | __main__:main:505 - Eval loss at step 2000: 3.836743302721095, Eval perplexity: 46.37420151254765 + Update steps: 20%|████▏ | 2001/10000 [03:58<12:09:42, 5.47s/it] Update steps: 20%|████▍ | 2003/10000 [03:58<8:33:34, 3.85s/it] Update steps: 20%|████▍ | 2005/10000 [03:58<6:02:21, 2.72s/it] Update steps: 20%|████▍ | 2007/10000 [03:58<4:16:25, 1.92s/it] Update steps: 20%|████▍ | 2009/10000 [03:58<3:02:17, 1.37s/it] Update steps: 20%|████▍ | 2011/10000 [03:59<2:10:24, 1.02it/s] Update steps: 20%|████▍ | 2013/10000 [03:59<1:34:09, 1.41it/s] Update steps: 20%|████▍ | 2015/10000 [03:59<1:08:46, 1.94it/s] Update steps: 20%|████▊ | 2017/10000 [03:59<50:59, 2.61it/s] Update steps: 20%|████▊ | 2019/10000 [03:59<38:38, 3.44it/s] Update steps: 20%|████▊ | 2021/10000 [03:59<29:54, 4.45it/s] Update steps: 20%|████▊ | 2023/10000 [03:59<23:46, 5.59it/s] Update steps: 20%|████▊ | 2025/10000 [04:00<19:29, 6.82it/s] Update steps: 20%|████▊ | 2027/10000 [04:00<16:33, 8.03it/s] Update steps: 20%|████▊ | 2029/10000 [04:00<14:28, 9.18it/s] Update steps: 20%|████▊ | 2031/10000 [04:00<13:00, 10.21it/s] Update steps: 20%|████▉ | 2033/10000 [04:00<12:05, 10.98it/s] Update steps: 20%|████▉ | 2035/10000 [04:00<11:19, 11.72it/s] Update steps: 20%|████▉ | 2037/10000 [04:00<10:50, 12.25it/s] Update steps: 20%|████▉ | 2039/10000 [04:01<10:30, 12.63it/s] Update steps: 20%|████▉ | 2041/10000 [04:01<10:13, 12.98it/s] Update steps: 20%|████▉ | 2043/10000 [04:01<10:02, 13.21it/s] Update steps: 20%|████▉ | 2045/10000 [04:01<09:54, 13.38it/s] Update steps: 20%|████▉ | 2047/10000 [04:01<09:49, 13.48it/s] Update steps: 20%|████▉ | 2049/10000 [04:01<09:51, 13.45it/s] Update steps: 21%|████▉ | 2051/10000 [04:01<09:43, 13.61it/s] Update steps: 21%|████▉ | 2053/10000 [04:02<09:38, 13.73it/s] Update steps: 21%|████▉ | 2055/10000 [04:02<09:39, 13.71it/s] Update steps: 21%|████▉ | 2057/10000 [04:02<09:37, 13.75it/s] Update steps: 21%|████▉ | 2059/10000 [04:02<09:37, 13.74it/s] Update steps: 21%|████▉ | 2061/10000 [04:02<09:40, 13.67it/s] Update steps: 21%|████▉ | 2063/10000 [04:02<09:37, 13.74it/s] Update steps: 21%|████▉ | 2065/10000 [04:02<09:36, 13.77it/s] Update steps: 21%|████▉ | 2067/10000 [04:03<09:43, 13.59it/s] Update steps: 21%|████▉ | 2069/10000 [04:03<09:41, 13.63it/s] Update steps: 21%|████▉ | 2071/10000 [04:03<09:37, 13.73it/s] Update steps: 21%|████▉ | 2073/10000 [04:03<09:34, 13.80it/s] Update steps: 21%|████▉ | 2075/10000 [04:03<09:31, 13.88it/s] Update steps: 21%|████▉ | 2077/10000 [04:03<09:32, 13.84it/s] Update steps: 21%|████▉ | 2079/10000 [04:04<09:39, 13.67it/s] Update steps: 21%|████▉ | 2081/10000 [04:04<09:38, 13.68it/s] Update steps: 21%|████▉ | 2083/10000 [04:04<09:31, 13.84it/s] Update steps: 21%|█████ | 2085/10000 [04:04<09:28, 13.92it/s] Update steps: 21%|█████ | 2087/10000 [04:04<09:26, 13.98it/s] Update steps: 21%|█████ | 2089/10000 [04:04<09:30, 13.86it/s] Update steps: 21%|█████ | 2091/10000 [04:04<09:31, 13.85it/s] Update steps: 21%|█████ | 2093/10000 [04:05<09:33, 13.80it/s] Update steps: 21%|█████ | 2095/10000 [04:05<09:36, 13.70it/s] Update steps: 21%|█████ | 2097/10000 [04:05<09:38, 13.66it/s] Update steps: 21%|█████ | 2099/10000 [04:05<09:32, 13.79it/s] Update steps: 21%|█████ | 2101/10000 [04:05<09:29, 13.87it/s] Update steps: 21%|█████ | 2103/10000 [04:05<09:26, 13.94it/s] Update steps: 21%|█████ | 2105/10000 [04:05<09:27, 13.92it/s] Update steps: 21%|█████ | 2107/10000 [04:06<09:26, 13.92it/s] Update steps: 21%|█████ | 2109/10000 [04:06<09:24, 13.97it/s] Update steps: 21%|█████ | 2111/10000 [04:06<09:26, 13.93it/s] Update steps: 21%|█████ | 2113/10000 [04:06<09:25, 13.96it/s] Update steps: 21%|█████ | 2115/10000 [04:06<09:32, 13.78it/s] Update steps: 21%|█████ | 2117/10000 [04:06<09:30, 13.83it/s] Update steps: 21%|█████ | 2119/10000 [04:06<09:29, 13.84it/s] Update steps: 21%|█████ | 2121/10000 [04:07<09:27, 13.87it/s] Update steps: 21%|█████ | 2123/10000 [04:07<09:29, 13.82it/s] Update steps: 21%|█████ | 2125/10000 [04:07<09:28, 13.86it/s] Update steps: 21%|█████ | 2127/10000 [04:07<09:34, 13.71it/s] Update steps: 21%|█████ | 2129/10000 [04:07<09:29, 13.82it/s] Update steps: 21%|█████ | 2131/10000 [04:07<09:28, 13.83it/s] Update steps: 21%|█████ | 2133/10000 [04:07<09:27, 13.87it/s] Update steps: 21%|█████ | 2135/10000 [04:08<09:32, 13.73it/s] Update steps: 21%|█████▏ | 2137/10000 [04:08<09:37, 13.62it/s] Update steps: 21%|█████▏ | 2139/10000 [04:08<09:32, 13.73it/s] Update steps: 21%|█████▏ | 2141/10000 [04:08<09:35, 13.66it/s] Update steps: 21%|█████▏ | 2143/10000 [04:08<09:29, 13.79it/s] Update steps: 21%|█████▏ | 2145/10000 [04:08<09:26, 13.86it/s] Update steps: 21%|█████▏ | 2147/10000 [04:08<09:24, 13.91it/s] Update steps: 21%|█████▏ | 2149/10000 [04:09<09:21, 13.98it/s] Update steps: 22%|█████▏ | 2151/10000 [04:09<09:25, 13.87it/s] Update steps: 22%|█████▏ | 2153/10000 [04:09<09:27, 13.84it/s] Update steps: 22%|█████▏ | 2155/10000 [04:09<09:25, 13.87it/s] Update steps: 22%|█████▏ | 2157/10000 [04:09<09:21, 13.96it/s] Update steps: 22%|█████▏ | 2159/10000 [04:09<09:21, 13.97it/s] Update steps: 22%|█████▏ | 2161/10000 [04:09<09:28, 13.80it/s] Update steps: 22%|█████▏ | 2163/10000 [04:10<09:28, 13.78it/s] Update steps: 22%|█████▏ | 2165/10000 [04:10<09:26, 13.84it/s] Update steps: 22%|█████▏ | 2167/10000 [04:10<09:23, 13.91it/s] Update steps: 22%|█████▏ | 2169/10000 [04:10<09:22, 13.93it/s] Update steps: 22%|█████▏ | 2171/10000 [04:10<09:25, 13.85it/s] Update steps: 22%|█████▏ | 2173/10000 [04:10<09:30, 13.73it/s] Update steps: 22%|█████▏ | 2175/10000 [04:10<09:31, 13.69it/s] Update steps: 22%|█████▏ | 2177/10000 [04:11<09:29, 13.74it/s] Update steps: 22%|█████▏ | 2179/10000 [04:11<09:26, 13.80it/s] Update steps: 22%|█████▏ | 2181/10000 [04:11<09:24, 13.85it/s] Update steps: 22%|█████▏ | 2183/10000 [04:11<09:25, 13.81it/s] Update steps: 22%|█████▏ | 2185/10000 [04:11<09:21, 13.91it/s] Update steps: 22%|█████▏ | 2187/10000 [04:11<09:25, 13.83it/s] Update steps: 22%|█████▎ | 2189/10000 [04:11<09:26, 13.80it/s] Update steps: 22%|█████▎ | 2191/10000 [04:12<09:24, 13.83it/s] Update steps: 22%|█████▎ | 2193/10000 [04:12<09:20, 13.93it/s] Update steps: 22%|█████▎ | 2195/10000 [04:12<09:25, 13.81it/s] Update steps: 22%|█████▎ | 2197/10000 [04:12<09:26, 13.78it/s] Update steps: 22%|█████▎ | 2199/10000 [04:12<09:22, 13.87it/s] Update steps: 22%|█████▎ | 2201/10000 [04:12<09:23, 13.85it/s] Update steps: 22%|█████▎ | 2203/10000 [04:12<09:21, 13.89it/s] Update steps: 22%|█████▎ | 2205/10000 [04:13<09:19, 13.94it/s] Update steps: 22%|█████▎ | 2207/10000 [04:13<09:20, 13.90it/s] Update steps: 22%|█████▎ | 2209/10000 [04:13<09:24, 13.81it/s] Update steps: 22%|█████▎ | 2211/10000 [04:13<09:24, 13.80it/s] Update steps: 22%|█████▎ | 2213/10000 [04:13<09:21, 13.88it/s] Update steps: 22%|█████▎ | 2215/10000 [04:13<09:18, 13.93it/s] Update steps: 22%|█████▎ | 2217/10000 [04:13<09:22, 13.83it/s] Update steps: 22%|█████▎ | 2219/10000 [04:14<09:24, 13.78it/s] Update steps: 22%|█████▎ | 2221/10000 [04:14<09:24, 13.77it/s] Update steps: 22%|█████▎ | 2223/10000 [04:14<09:25, 13.76it/s] Update steps: 22%|█████▎ | 2225/10000 [04:14<09:25, 13.75it/s] Update steps: 22%|█████▎ | 2227/10000 [04:14<09:28, 13.67it/s] Update steps: 22%|█████▎ | 2229/10000 [04:14<09:25, 13.75it/s] Update steps: 22%|█████▎ | 2231/10000 [04:14<09:26, 13.72it/s] Update steps: 22%|█████▎ | 2233/10000 [04:15<09:22, 13.81it/s] Update steps: 22%|█████▎ | 2235/10000 [04:15<09:23, 13.77it/s] Update steps: 22%|█████▎ | 2237/10000 [04:15<09:20, 13.85it/s] Update steps: 22%|█████▎ | 2239/10000 [04:15<09:21, 13.83it/s] Update steps: 22%|█████▍ | 2241/10000 [04:15<09:22, 13.80it/s] Update steps: 22%|█████▍ | 2243/10000 [04:15<09:21, 13.82it/s] Update steps: 22%|█████▍ | 2245/10000 [04:16<09:19, 13.85it/s] Update steps: 22%|█████▍ | 2247/10000 [04:16<09:20, 13.82it/s] Update steps: 22%|█████▍ | 2249/10000 [04:16<09:18, 13.88it/s] Update steps: 23%|█████▍ | 2251/10000 [04:16<09:20, 13.82it/s] Update steps: 23%|█████▍ | 2253/10000 [04:16<09:25, 13.70it/s] Update steps: 23%|█████▍ | 2255/10000 [04:16<09:30, 13.58it/s] Update steps: 23%|█████▍ | 2257/10000 [04:16<09:30, 13.56it/s] Update steps: 23%|█████▍ | 2259/10000 [04:17<09:26, 13.66it/s] Update steps: 23%|█████▍ | 2261/10000 [04:17<09:27, 13.64it/s] Update steps: 23%|█████▍ | 2263/10000 [04:17<09:24, 13.71it/s] Update steps: 23%|█████▍ | 2265/10000 [04:17<09:27, 13.63it/s] Update steps: 23%|█████▍ | 2267/10000 [04:17<09:30, 13.56it/s] Update steps: 23%|█████▍ | 2269/10000 [04:17<09:24, 13.69it/s] Update steps: 23%|█████▍ | 2271/10000 [04:17<09:20, 13.80it/s] Update steps: 23%|█████▍ | 2273/10000 [04:18<09:24, 13.68it/s] Update steps: 23%|█████▍ | 2275/10000 [04:18<09:26, 13.64it/s] Update steps: 23%|█████▍ | 2277/10000 [04:18<09:27, 13.62it/s] Update steps: 23%|█████▍ | 2279/10000 [04:18<09:22, 13.74it/s] Update steps: 23%|█████▍ | 2281/10000 [04:18<09:28, 13.58it/s] Update steps: 23%|█████▍ | 2283/10000 [04:18<09:25, 13.65it/s] Update steps: 23%|█████▍ | 2285/10000 [04:18<09:30, 13.52it/s] Update steps: 23%|█████▍ | 2287/10000 [04:19<09:34, 13.42it/s] Update steps: 23%|█████▍ | 2289/10000 [04:19<09:27, 13.58it/s] Update steps: 23%|█████▍ | 2291/10000 [04:19<09:20, 13.75it/s] Update steps: 23%|█████▌ | 2293/10000 [04:19<09:21, 13.71it/s] Update steps: 23%|█████▌ | 2295/10000 [04:19<09:18, 13.80it/s] Update steps: 23%|█████▌ | 2297/10000 [04:19<09:16, 13.84it/s] Update steps: 23%|█████▌ | 2299/10000 [04:19<09:23, 13.67it/s] Update steps: 23%|█████▌ | 2301/10000 [04:20<09:18, 13.78it/s] Update steps: 23%|█████▌ | 2303/10000 [04:20<09:16, 13.83it/s] Update steps: 23%|█████▌ | 2305/10000 [04:20<09:21, 13.71it/s] Update steps: 23%|█████▌ | 2307/10000 [04:20<09:45, 13.13it/s] Update steps: 23%|█████▌ | 2309/10000 [04:20<09:39, 13.26it/s] Update steps: 23%|█████▌ | 2311/10000 [04:20<09:30, 13.48it/s] Update steps: 23%|█████▌ | 2313/10000 [04:20<09:27, 13.54it/s] Update steps: 23%|█████▌ | 2315/10000 [04:21<09:31, 13.44it/s] Update steps: 23%|█████▌ | 2317/10000 [04:21<09:27, 13.55it/s] Update steps: 23%|█████▌ | 2319/10000 [04:21<09:23, 13.62it/s] Update steps: 23%|█████▌ | 2321/10000 [04:21<09:18, 13.74it/s] Update steps: 23%|█████▌ | 2323/10000 [04:21<09:19, 13.72it/s] Update steps: 23%|█████▌ | 2325/10000 [04:21<09:17, 13.76it/s] Update steps: 23%|█████▌ | 2327/10000 [04:22<09:22, 13.64it/s] Update steps: 23%|█████▌ | 2329/10000 [04:22<09:17, 13.75it/s] Update steps: 23%|█████▌ | 2331/10000 [04:22<09:14, 13.83it/s] Update steps: 23%|█████▌ | 2333/10000 [04:22<09:19, 13.71it/s] Update steps: 23%|█████▌ | 2335/10000 [04:22<09:17, 13.74it/s] Update steps: 23%|█████▌ | 2337/10000 [04:22<09:15, 13.80it/s] Update steps: 23%|█████▌ | 2339/10000 [04:22<09:14, 13.82it/s] Update steps: 23%|█████▌ | 2341/10000 [04:23<09:21, 13.64it/s] Update steps: 23%|█████▌ | 2343/10000 [04:23<09:23, 13.58it/s] Update steps: 23%|█████▋ | 2345/10000 [04:23<09:20, 13.67it/s] Update steps: 23%|█████▋ | 2347/10000 [04:23<09:21, 13.62it/s] Update steps: 23%|█████▋ | 2349/10000 [04:23<09:23, 13.58it/s] Update steps: 24%|█████▋ | 2351/10000 [04:23<09:25, 13.51it/s] Update steps: 24%|█████▋ | 2353/10000 [04:23<09:25, 13.52it/s] Update steps: 24%|█████▋ | 2355/10000 [04:24<09:25, 13.51it/s] Update steps: 24%|█████▋ | 2357/10000 [04:24<09:27, 13.46it/s] Update steps: 24%|█████▋ | 2359/10000 [04:24<09:20, 13.62it/s] Update steps: 24%|█████▋ | 2361/10000 [04:24<09:17, 13.71it/s] Update steps: 24%|█████▋ | 2363/10000 [04:24<09:17, 13.70it/s] Update steps: 24%|█████▋ | 2365/10000 [04:24<09:13, 13.79it/s] Update steps: 24%|█████▋ | 2367/10000 [04:24<09:13, 13.80it/s] Update steps: 24%|█████▋ | 2369/10000 [04:25<09:13, 13.78it/s] Update steps: 24%|█████▋ | 2371/10000 [04:25<09:15, 13.74it/s] Update steps: 24%|█████▋ | 2373/10000 [04:25<09:19, 13.62it/s] Update steps: 24%|█████▋ | 2375/10000 [04:25<09:24, 13.51it/s] Update steps: 24%|█████▋ | 2377/10000 [04:25<09:24, 13.52it/s] Update steps: 24%|█████▋ | 2379/10000 [04:25<09:19, 13.63it/s] Update steps: 24%|█████▋ | 2381/10000 [04:25<09:14, 13.73it/s] Update steps: 24%|█████▋ | 2383/10000 [04:26<09:13, 13.77it/s] Update steps: 24%|█████▋ | 2385/10000 [04:26<09:12, 13.78it/s] Update steps: 24%|█████▋ | 2387/10000 [04:26<09:28, 13.39it/s] Update steps: 24%|█████▋ | 2389/10000 [04:26<09:32, 13.28it/s] Update steps: 24%|█████▋ | 2391/10000 [04:26<09:29, 13.36it/s] Update steps: 24%|█████▋ | 2393/10000 [04:26<09:29, 13.35it/s] Update steps: 24%|█████▋ | 2395/10000 [04:27<09:22, 13.53it/s] Update steps: 24%|█████▊ | 2397/10000 [04:27<09:15, 13.69it/s] Update steps: 24%|█████▊ | 2399/10000 [04:27<09:11, 13.78it/s] Update steps: 24%|█████▊ | 2401/10000 [04:27<09:11, 13.78it/s] Update steps: 24%|█████▊ | 2403/10000 [04:27<09:13, 13.73it/s] Update steps: 24%|█████▊ | 2405/10000 [04:27<09:08, 13.85it/s] Update steps: 24%|█████▊ | 2407/10000 [04:27<09:08, 13.85it/s] Update steps: 24%|█████▊ | 2409/10000 [04:28<09:15, 13.66it/s] Update steps: 24%|█████▊ | 2411/10000 [04:28<09:14, 13.69it/s] Update steps: 24%|█████▊ | 2413/10000 [04:28<09:18, 13.58it/s] Update steps: 24%|█████▊ | 2415/10000 [04:28<09:34, 13.20it/s] Update steps: 24%|█████▊ | 2417/10000 [04:28<09:32, 13.24it/s] Update steps: 24%|█████▊ | 2419/10000 [04:28<09:23, 13.45it/s] Update steps: 24%|█████▊ | 2421/10000 [04:28<09:21, 13.50it/s] Update steps: 24%|█████▊ | 2423/10000 [04:29<09:18, 13.57it/s] Update steps: 24%|█████▊ | 2425/10000 [04:29<09:17, 13.60it/s] Update steps: 24%|█████▊ | 2427/10000 [04:29<09:14, 13.66it/s] Update steps: 24%|█████▊ | 2429/10000 [04:29<09:14, 13.66it/s] Update steps: 24%|█████▊ | 2431/10000 [04:29<09:11, 13.73it/s] Update steps: 24%|█████▊ | 2433/10000 [04:29<09:10, 13.75it/s] Update steps: 24%|█████▊ | 2435/10000 [04:29<09:16, 13.60it/s] Update steps: 24%|█████▊ | 2437/10000 [04:30<09:22, 13.44it/s] Update steps: 24%|█████▊ | 2439/10000 [04:30<09:20, 13.49it/s] Update steps: 24%|█████▊ | 2441/10000 [04:30<09:15, 13.62it/s] Update steps: 24%|█████▊ | 2443/10000 [04:30<09:16, 13.59it/s] Update steps: 24%|█████▊ | 2445/10000 [04:30<09:18, 13.53it/s] Update steps: 24%|█████▊ | 2447/10000 [04:30<09:14, 13.63it/s] Update steps: 24%|█████▉ | 2449/10000 [04:30<09:16, 13.57it/s] Update steps: 25%|█████▉ | 2451/10000 [04:31<09:14, 13.62it/s] Update steps: 25%|█████▉ | 2453/10000 [04:31<09:14, 13.61it/s] Update steps: 25%|█████▉ | 2455/10000 [04:31<09:11, 13.68it/s] Update steps: 25%|█████▉ | 2457/10000 [04:31<09:07, 13.78it/s] Update steps: 25%|█████▉ | 2459/10000 [04:31<09:06, 13.80it/s] Update steps: 25%|█████▉ | 2461/10000 [04:31<09:05, 13.83it/s] Update steps: 25%|█████▉ | 2463/10000 [04:31<09:06, 13.80it/s] Update steps: 25%|█████▉ | 2465/10000 [04:32<09:11, 13.67it/s] Update steps: 25%|█████▉ | 2467/10000 [04:32<09:13, 13.61it/s] Update steps: 25%|█████▉ | 2469/10000 [04:32<09:09, 13.72it/s] Update steps: 25%|█████▉ | 2471/10000 [04:32<09:08, 13.72it/s] Update steps: 25%|█████▉ | 2473/10000 [04:32<09:08, 13.73it/s] Update steps: 25%|█████▉ | 2475/10000 [04:32<09:12, 13.61it/s] Update steps: 25%|█████▉ | 2477/10000 [04:33<09:08, 13.70it/s] Update steps: 25%|█████▉ | 2479/10000 [04:33<09:07, 13.74it/s] Update steps: 25%|█████▉ | 2481/10000 [04:33<09:07, 13.73it/s] Update steps: 25%|█████▉ | 2483/10000 [04:33<09:05, 13.78it/s] Update steps: 25%|█████▉ | 2485/10000 [04:33<09:10, 13.65it/s] Update steps: 25%|█████▉ | 2487/10000 [04:33<09:13, 13.58it/s] Update steps: 25%|█████▉ | 2489/10000 [04:33<09:10, 13.65it/s] Update steps: 25%|█████▉ | 2491/10000 [04:34<09:04, 13.78it/s] Update steps: 25%|█████▉ | 2493/10000 [04:34<09:04, 13.78it/s] Update steps: 25%|█████▉ | 2495/10000 [04:34<09:06, 13.73it/s] Update steps: 25%|█████▉ | 2497/10000 [04:34<09:13, 13.56it/s] Update steps: 25%|█████▉ | 2499/10000 [04:34<09:10, 13.63it/s] Update steps: 25%|██████ | 2501/10000 [04:34<09:05, 13.76it/s] Update steps: 25%|██████ | 2503/10000 [04:34<09:13, 13.54it/s] Update steps: 25%|██████ | 2505/10000 [04:35<09:07, 13.69it/s] Update steps: 25%|██████ | 2507/10000 [04:35<09:05, 13.73it/s] Update steps: 25%|██████ | 2509/10000 [04:35<09:09, 13.62it/s] Update steps: 25%|██████ | 2511/10000 [04:35<09:08, 13.65it/s] Update steps: 25%|██████ | 2513/10000 [04:35<09:04, 13.75it/s] Update steps: 25%|██████ | 2515/10000 [04:35<09:04, 13.75it/s] Update steps: 25%|██████ | 2517/10000 [04:35<09:02, 13.80it/s] Update steps: 25%|██████ | 2519/10000 [04:36<08:59, 13.86it/s] Update steps: 25%|██████ | 2521/10000 [04:36<08:58, 13.90it/s] Update steps: 25%|██████ | 2523/10000 [04:36<08:57, 13.92it/s] Update steps: 25%|██████ | 2525/10000 [04:36<09:00, 13.84it/s] Update steps: 25%|██████ | 2527/10000 [04:36<09:09, 13.59it/s] Update steps: 25%|██████ | 2529/10000 [04:36<09:07, 13.65it/s] Update steps: 25%|██████ | 2531/10000 [04:36<09:06, 13.67it/s] Update steps: 25%|██████ | 2533/10000 [04:37<09:09, 13.59it/s] Update steps: 25%|██████ | 2535/10000 [04:37<09:10, 13.55it/s] Update steps: 25%|██████ | 2537/10000 [04:37<09:06, 13.65it/s] Update steps: 25%|██████ | 2539/10000 [04:37<09:02, 13.75it/s] Update steps: 25%|██████ | 2541/10000 [04:37<09:02, 13.74it/s] Update steps: 25%|██████ | 2543/10000 [04:37<09:02, 13.75it/s] Update steps: 25%|██████ | 2545/10000 [04:37<09:03, 13.73it/s] Update steps: 25%|██████ | 2547/10000 [04:38<09:01, 13.76it/s] Update steps: 25%|██████ | 2549/10000 [04:38<09:05, 13.65it/s] Update steps: 26%|██████ | 2551/10000 [04:38<09:03, 13.70it/s] Update steps: 26%|██████▏ | 2553/10000 [04:38<08:59, 13.81it/s] Update steps: 26%|██████▏ | 2555/10000 [04:38<08:57, 13.86it/s] Update steps: 26%|██████▏ | 2557/10000 [04:38<08:57, 13.86it/s] Update steps: 26%|██████▏ | 2559/10000 [04:38<09:03, 13.68it/s] Update steps: 26%|██████▏ | 2561/10000 [04:39<09:02, 13.70it/s] Update steps: 26%|██████▏ | 2563/10000 [04:39<09:03, 13.68it/s] Update steps: 26%|██████▏ | 2565/10000 [04:39<09:10, 13.51it/s] Update steps: 26%|██████▏ | 2567/10000 [04:39<09:05, 13.63it/s] Update steps: 26%|██████▏ | 2569/10000 [04:39<09:00, 13.75it/s] Update steps: 26%|██████▏ | 2571/10000 [04:39<09:00, 13.74it/s] Update steps: 26%|██████▏ | 2573/10000 [04:40<08:59, 13.76it/s] Update steps: 26%|██████▏ | 2575/10000 [04:40<09:01, 13.71it/s] Update steps: 26%|██████▏ | 2577/10000 [04:40<09:00, 13.74it/s] Update steps: 26%|██████▏ | 2579/10000 [04:40<09:04, 13.64it/s] Update steps: 26%|██████▏ | 2581/10000 [04:40<09:03, 13.65it/s] Update steps: 26%|██████▏ | 2583/10000 [04:40<08:58, 13.78it/s] Update steps: 26%|██████▏ | 2585/10000 [04:40<09:01, 13.69it/s] Update steps: 26%|██████▏ | 2587/10000 [04:41<08:58, 13.76it/s] Update steps: 26%|██████▏ | 2589/10000 [04:41<09:00, 13.70it/s] Update steps: 26%|██████▏ | 2591/10000 [04:41<09:03, 13.64it/s] Update steps: 26%|██████▏ | 2593/10000 [04:41<08:57, 13.78it/s] Update steps: 26%|██████▏ | 2595/10000 [04:41<08:55, 13.82it/s] Update steps: 26%|██████▏ | 2597/10000 [04:41<08:58, 13.75it/s] Update steps: 26%|██████▏ | 2599/10000 [04:41<08:55, 13.83it/s] Update steps: 26%|██████▏ | 2601/10000 [04:42<08:54, 13.85it/s] Update steps: 26%|██████▏ | 2603/10000 [04:42<08:58, 13.74it/s] Update steps: 26%|██████▎ | 2605/10000 [04:42<09:01, 13.66it/s] Update steps: 26%|██████▎ | 2607/10000 [04:42<09:10, 13.43it/s] Update steps: 26%|██████▎ | 2609/10000 [04:42<09:11, 13.40it/s] Update steps: 26%|██████▎ | 2611/10000 [04:42<09:04, 13.58it/s] Update steps: 26%|██████▎ | 2613/10000 [04:42<09:00, 13.67it/s] Update steps: 26%|██████▎ | 2615/10000 [04:43<08:58, 13.71it/s] Update steps: 26%|██████▎ | 2617/10000 [04:43<08:57, 13.73it/s] Update steps: 26%|██████▎ | 2619/10000 [04:43<08:58, 13.70it/s] Update steps: 26%|██████▎ | 2621/10000 [04:43<08:55, 13.77it/s] Update steps: 26%|██████▎ | 2623/10000 [04:43<08:53, 13.84it/s] Update steps: 26%|██████▎ | 2625/10000 [04:43<08:52, 13.86it/s] Update steps: 26%|██████▎ | 2627/10000 [04:43<08:52, 13.86it/s] Update steps: 26%|██████▎ | 2629/10000 [04:44<08:55, 13.77it/s] Update steps: 26%|██████▎ | 2631/10000 [04:44<09:00, 13.63it/s] Update steps: 26%|██████▎ | 2633/10000 [04:44<09:09, 13.40it/s] Update steps: 26%|██████▎ | 2635/10000 [04:44<09:08, 13.43it/s] Update steps: 26%|██████▎ | 2637/10000 [04:44<09:10, 13.37it/s] Update steps: 26%|██████▎ | 2639/10000 [04:44<10:35, 11.59it/s] Update steps: 26%|██████▎ | 2641/10000 [04:45<10:32, 11.63it/s] Update steps: 26%|██████▎ | 2643/10000 [04:45<10:00, 12.26it/s] Update steps: 26%|██████▎ | 2645/10000 [04:45<09:38, 12.70it/s] Update steps: 26%|██████▎ | 2647/10000 [04:45<09:25, 13.01it/s] Update steps: 26%|██████▎ | 2649/10000 [04:45<09:19, 13.14it/s] Update steps: 27%|██████▎ | 2651/10000 [04:45<09:17, 13.19it/s] Update steps: 27%|██████▎ | 2653/10000 [04:45<09:06, 13.43it/s] Update steps: 27%|██████▎ | 2655/10000 [04:46<09:04, 13.50it/s] Update steps: 27%|██████▍ | 2657/10000 [04:46<09:03, 13.52it/s] Update steps: 27%|██████▍ | 2659/10000 [04:46<09:05, 13.46it/s] Update steps: 27%|██████▍ | 2661/10000 [04:46<09:02, 13.52it/s] Update steps: 27%|██████▍ | 2663/10000 [04:46<08:59, 13.61it/s] Update steps: 27%|██████▍ | 2665/10000 [04:46<08:57, 13.65it/s] Update steps: 27%|██████▍ | 2667/10000 [04:47<09:00, 13.56it/s] Update steps: 27%|██████▍ | 2669/10000 [04:47<08:59, 13.58it/s] Update steps: 27%|██████▍ | 2671/10000 [04:47<08:54, 13.71it/s] Update steps: 27%|██████▍ | 2673/10000 [04:47<08:55, 13.69it/s] Update steps: 27%|██████▍ | 2675/10000 [04:47<08:59, 13.58it/s] Update steps: 27%|██████▍ | 2677/10000 [04:47<08:55, 13.67it/s] Update steps: 27%|██████▍ | 2679/10000 [04:47<08:53, 13.73it/s] Update steps: 27%|██████▍ | 2681/10000 [04:48<08:48, 13.85it/s] Update steps: 27%|██████▍ | 2683/10000 [04:48<08:49, 13.83it/s] Update steps: 27%|██████▍ | 2685/10000 [04:48<08:45, 13.92it/s] Update steps: 27%|██████▍ | 2687/10000 [04:48<08:44, 13.94it/s] Update steps: 27%|██████▍ | 2689/10000 [04:48<08:48, 13.82it/s] Update steps: 27%|██████▍ | 2691/10000 [04:48<08:46, 13.89it/s] Update steps: 27%|██████▍ | 2693/10000 [04:48<08:45, 13.90it/s] Update steps: 27%|██████▍ | 2695/10000 [04:49<08:44, 13.92it/s] Update steps: 27%|██████▍ | 2697/10000 [04:49<08:44, 13.93it/s] Update steps: 27%|██████▍ | 2699/10000 [04:49<08:47, 13.84it/s] Update steps: 27%|██████▍ | 2701/10000 [04:49<08:48, 13.80it/s] Update steps: 27%|██████▍ | 2703/10000 [04:49<08:47, 13.84it/s] Update steps: 27%|██████▍ | 2705/10000 [04:49<08:56, 13.61it/s] Update steps: 27%|██████▍ | 2707/10000 [04:49<08:53, 13.66it/s] Update steps: 27%|██████▌ | 2709/10000 [04:50<08:50, 13.74it/s] Update steps: 27%|██████▌ | 2711/10000 [04:50<08:49, 13.78it/s] Update steps: 27%|██████▌ | 2713/10000 [04:50<08:50, 13.75it/s] Update steps: 27%|██████▌ | 2715/10000 [04:50<08:50, 13.73it/s] Update steps: 27%|██████▌ | 2717/10000 [04:50<08:48, 13.79it/s] Update steps: 27%|██████▌ | 2719/10000 [04:50<08:52, 13.68it/s] Update steps: 27%|██████▌ | 2721/10000 [04:50<08:54, 13.62it/s] Update steps: 27%|██████▌ | 2723/10000 [04:51<08:49, 13.75it/s] Update steps: 27%|██████▌ | 2725/10000 [04:51<08:50, 13.73it/s] Update steps: 27%|██████▌ | 2727/10000 [04:51<08:48, 13.77it/s] Update steps: 27%|██████▌ | 2729/10000 [04:51<08:49, 13.73it/s] Update steps: 27%|██████▌ | 2731/10000 [04:51<08:51, 13.68it/s] Update steps: 27%|██████▌ | 2733/10000 [04:51<08:53, 13.61it/s] Update steps: 27%|██████▌ | 2735/10000 [04:51<09:07, 13.27it/s] Update steps: 27%|██████▌ | 2737/10000 [04:52<08:59, 13.47it/s] Update steps: 27%|██████▌ | 2739/10000 [04:52<09:00, 13.44it/s] Update steps: 27%|██████▌ | 2741/10000 [04:52<08:54, 13.59it/s] Update steps: 27%|██████▌ | 2743/10000 [04:52<08:59, 13.44it/s] Update steps: 27%|██████▌ | 2745/10000 [04:52<08:54, 13.58it/s] Update steps: 27%|██████▌ | 2747/10000 [04:52<08:52, 13.62it/s] Update steps: 27%|██████▌ | 2749/10000 [04:52<08:52, 13.62it/s] Update steps: 28%|██████▌ | 2751/10000 [04:53<08:55, 13.54it/s] Update steps: 28%|██████▌ | 2753/10000 [04:53<09:08, 13.22it/s] Update steps: 28%|██████▌ | 2755/10000 [04:53<09:08, 13.21it/s] Update steps: 28%|██████▌ | 2757/10000 [04:53<09:00, 13.40it/s] Update steps: 28%|██████▌ | 2759/10000 [04:53<09:03, 13.31it/s] Update steps: 28%|██████▋ | 2761/10000 [04:53<09:00, 13.39it/s] Update steps: 28%|██████▋ | 2763/10000 [04:54<08:55, 13.51it/s] Update steps: 28%|██████▋ | 2765/10000 [04:54<08:59, 13.40it/s] Update steps: 28%|██████▋ | 2767/10000 [04:54<08:55, 13.50it/s] Update steps: 28%|██████▋ | 2769/10000 [04:54<08:56, 13.47it/s] Update steps: 28%|██████▋ | 2771/10000 [04:54<08:54, 13.51it/s] Update steps: 28%|██████▋ | 2773/10000 [04:54<08:52, 13.57it/s] Update steps: 28%|██████▋ | 2775/10000 [04:54<08:49, 13.65it/s] Update steps: 28%|██████▋ | 2777/10000 [04:55<08:49, 13.64it/s] Update steps: 28%|██████▋ | 2779/10000 [04:55<08:46, 13.70it/s] Update steps: 28%|██████▋ | 2781/10000 [04:55<08:56, 13.45it/s] Update steps: 28%|██████▋ | 2783/10000 [04:55<08:50, 13.61it/s] Update steps: 28%|██████▋ | 2785/10000 [04:55<08:54, 13.50it/s] Update steps: 28%|██████▋ | 2787/10000 [04:55<08:53, 13.51it/s] Update steps: 28%|██████▋ | 2789/10000 [04:55<08:55, 13.46it/s] Update steps: 28%|██████▋ | 2791/10000 [04:56<08:49, 13.62it/s] Update steps: 28%|██████▋ | 2793/10000 [04:56<08:45, 13.72it/s] Update steps: 28%|██████▋ | 2795/10000 [04:56<08:47, 13.65it/s] Update steps: 28%|██████▋ | 2797/10000 [04:56<08:45, 13.72it/s] Update steps: 28%|██████▋ | 2799/10000 [04:56<08:44, 13.74it/s] Update steps: 28%|██████▋ | 2801/10000 [04:56<08:48, 13.61it/s] Update steps: 28%|██████▋ | 2803/10000 [04:56<08:48, 13.61it/s] Update steps: 28%|██████▋ | 2805/10000 [04:57<08:43, 13.73it/s] Update steps: 28%|██████▋ | 2807/10000 [04:57<08:42, 13.78it/s] Update steps: 28%|██████▋ | 2809/10000 [04:57<08:45, 13.67it/s] Update steps: 28%|██████▋ | 2811/10000 [04:57<08:48, 13.60it/s] Update steps: 28%|██████▊ | 2813/10000 [04:57<08:46, 13.65it/s] Update steps: 28%|██████▊ | 2815/10000 [04:57<08:55, 13.42it/s] Update steps: 28%|██████▊ | 2817/10000 [04:58<08:50, 13.55it/s] Update steps: 28%|██████▊ | 2819/10000 [04:58<08:48, 13.59it/s] Update steps: 28%|██████▊ | 2821/10000 [04:58<08:48, 13.58it/s] Update steps: 28%|██████▊ | 2823/10000 [04:58<08:44, 13.68it/s] Update steps: 28%|██████▊ | 2825/10000 [04:58<08:54, 13.42it/s] Update steps: 28%|██████▊ | 2827/10000 [04:58<08:56, 13.37it/s] Update steps: 28%|██████▊ | 2829/10000 [04:58<08:54, 13.42it/s] Update steps: 28%|██████▊ | 2831/10000 [04:59<08:47, 13.58it/s] Update steps: 28%|██████▊ | 2833/10000 [04:59<08:46, 13.62it/s] Update steps: 28%|██████▊ | 2835/10000 [04:59<08:47, 13.58it/s] Update steps: 28%|██████▊ | 2837/10000 [04:59<08:52, 13.46it/s] Update steps: 28%|██████▊ | 2839/10000 [04:59<09:00, 13.25it/s] Update steps: 28%|██████▊ | 2841/10000 [04:59<08:51, 13.47it/s] Update steps: 28%|██████▊ | 2843/10000 [04:59<08:46, 13.59it/s] Update steps: 28%|██████▊ | 2845/10000 [05:00<08:48, 13.54it/s] Update steps: 28%|██████▊ | 2847/10000 [05:00<08:44, 13.64it/s] Update steps: 28%|██████▊ | 2849/10000 [05:00<08:46, 13.58it/s] Update steps: 29%|██████▊ | 2851/10000 [05:00<08:42, 13.67it/s] Update steps: 29%|██████▊ | 2853/10000 [05:00<08:43, 13.64it/s] Update steps: 29%|██████▊ | 2855/10000 [05:00<08:43, 13.64it/s] Update steps: 29%|██████▊ | 2857/10000 [05:00<08:41, 13.71it/s] Update steps: 29%|██████▊ | 2859/10000 [05:01<08:40, 13.72it/s] Update steps: 29%|██████▊ | 2861/10000 [05:01<08:40, 13.72it/s] Update steps: 29%|██████▊ | 2863/10000 [05:01<08:41, 13.69it/s] Update steps: 29%|██████▉ | 2865/10000 [05:01<08:42, 13.66it/s] Update steps: 29%|██████▉ | 2867/10000 [05:01<08:42, 13.65it/s] Update steps: 29%|██████▉ | 2869/10000 [05:01<08:47, 13.51it/s] Update steps: 29%|██████▉ | 2871/10000 [05:01<08:42, 13.64it/s] Update steps: 29%|██████▉ | 2873/10000 [05:02<08:42, 13.64it/s] Update steps: 29%|██████▉ | 2875/10000 [05:02<08:44, 13.59it/s] Update steps: 29%|██████▉ | 2877/10000 [05:02<08:39, 13.72it/s] Update steps: 29%|██████▉ | 2879/10000 [05:02<08:35, 13.81it/s] Update steps: 29%|██████▉ | 2881/10000 [05:02<08:34, 13.82it/s] Update steps: 29%|██████▉ | 2883/10000 [05:02<08:34, 13.82it/s] Update steps: 29%|██████▉ | 2885/10000 [05:02<08:35, 13.81it/s] Update steps: 29%|██████▉ | 2887/10000 [05:03<08:33, 13.84it/s] Update steps: 29%|██████▉ | 2889/10000 [05:03<08:43, 13.58it/s] Update steps: 29%|██████▉ | 2891/10000 [05:03<08:43, 13.58it/s] Update steps: 29%|██████▉ | 2893/10000 [05:03<08:41, 13.62it/s] Update steps: 29%|██████▉ | 2895/10000 [05:03<08:46, 13.50it/s] Update steps: 29%|██████▉ | 2897/10000 [05:03<08:43, 13.56it/s] Update steps: 29%|██████▉ | 2899/10000 [05:04<08:44, 13.53it/s] Update steps: 29%|██████▉ | 2901/10000 [05:04<08:39, 13.68it/s] Update steps: 29%|██████▉ | 2903/10000 [05:04<08:42, 13.59it/s] Update steps: 29%|██████▉ | 2905/10000 [05:04<08:42, 13.57it/s] Update steps: 29%|██████▉ | 2907/10000 [05:04<08:40, 13.62it/s] Update steps: 29%|██████▉ | 2909/10000 [05:04<08:37, 13.72it/s] Update steps: 29%|██████▉ | 2911/10000 [05:04<08:39, 13.63it/s] Update steps: 29%|██████▉ | 2913/10000 [05:05<08:35, 13.74it/s] Update steps: 29%|██████▉ | 2915/10000 [05:05<08:33, 13.79it/s] Update steps: 29%|███████ | 2917/10000 [05:05<08:32, 13.83it/s] Update steps: 29%|███████ | 2919/10000 [05:05<08:33, 13.79it/s] Update steps: 29%|███████ | 2921/10000 [05:05<08:38, 13.66it/s] Update steps: 29%|███████ | 2923/10000 [05:05<08:38, 13.64it/s] Update steps: 29%|███████ | 2925/10000 [05:05<08:40, 13.60it/s] Update steps: 29%|███████ | 2927/10000 [05:06<08:49, 13.36it/s] Update steps: 29%|███████ | 2929/10000 [05:06<08:42, 13.53it/s] Update steps: 29%|███████ | 2931/10000 [05:06<08:40, 13.59it/s] Update steps: 29%|███████ | 2933/10000 [05:06<08:40, 13.59it/s] Update steps: 29%|███████ | 2935/10000 [05:06<08:47, 13.38it/s] Update steps: 29%|███████ | 2937/10000 [05:06<08:41, 13.55it/s] Update steps: 29%|███████ | 2939/10000 [05:06<08:38, 13.62it/s] Update steps: 29%|███████ | 2941/10000 [05:07<08:39, 13.59it/s] Update steps: 29%|███████ | 2943/10000 [05:07<08:40, 13.57it/s] Update steps: 29%|███████ | 2945/10000 [05:07<08:39, 13.57it/s] Update steps: 29%|███████ | 2947/10000 [05:07<08:38, 13.60it/s] Update steps: 29%|███████ | 2949/10000 [05:07<08:34, 13.71it/s] Update steps: 30%|███████ | 2951/10000 [05:07<08:34, 13.71it/s] Update steps: 30%|███████ | 2953/10000 [05:07<08:36, 13.64it/s] Update steps: 30%|███████ | 2955/10000 [05:08<08:36, 13.65it/s] Update steps: 30%|███████ | 2957/10000 [05:08<08:39, 13.55it/s] Update steps: 30%|███████ | 2959/10000 [05:08<08:35, 13.66it/s] Update steps: 30%|███████ | 2961/10000 [05:08<08:37, 13.61it/s] Update steps: 30%|███████ | 2963/10000 [05:08<08:38, 13.57it/s] Update steps: 30%|███████ | 2965/10000 [05:08<08:40, 13.50it/s] Update steps: 30%|███████ | 2967/10000 [05:09<08:38, 13.56it/s] Update steps: 30%|███████▏ | 2969/10000 [05:09<08:41, 13.48it/s] Update steps: 30%|███████▏ | 2971/10000 [05:09<08:39, 13.53it/s] Update steps: 30%|███████▏ | 2973/10000 [05:09<08:35, 13.62it/s] Update steps: 30%|███████▏ | 2975/10000 [05:09<08:32, 13.71it/s] Update steps: 30%|███████▏ | 2977/10000 [05:09<08:28, 13.82it/s] Update steps: 30%|███████▏ | 2979/10000 [05:09<08:32, 13.70it/s] Update steps: 30%|███████▏ | 2981/10000 [05:10<08:28, 13.80it/s] Update steps: 30%|███████▏ | 2983/10000 [05:10<08:31, 13.71it/s] Update steps: 30%|███████▏ | 2985/10000 [05:10<08:34, 13.63it/s] Update steps: 30%|███████▏ | 2987/10000 [05:10<08:35, 13.60it/s] Update steps: 30%|███████▏ | 2989/10000 [05:10<08:30, 13.73it/s] Update steps: 30%|███████▏ | 2991/10000 [05:10<08:32, 13.67it/s] Update steps: 30%|███████▏ | 2993/10000 [05:10<08:32, 13.66it/s] Update steps: 30%|███████▏ | 2995/10000 [05:11<08:33, 13.64it/s] Update steps: 30%|███████▏ | 2997/10000 [05:11<08:34, 13.62it/s] Update steps: 30%|███████▏ | 2999/10000 [05:11<08:29, 13.73it/s]2026-04-17 10:58:31.831 | INFO | __main__:main:490 - Performing evaluation at step 3000 +2026-04-17 10:58:31.833 | INFO | utils.eval:evaluate_model:21 - Loading validation data from 8 local validation files +2026-04-17 10:58:32.578 | INFO | utils.eval:evaluate_model:43 - Loaded validation dataset in 0.75 seconds +2026-04-17 10:58:32.579 | INFO | utils.eval:evaluate_model:60 - Eval set prepared in 0.75 seconds + Update steps: 30%|███████▏ | 3000/10000 [05:26<08:29, 13.73it/s]2026-04-17 10:59:07.641 | INFO | __main__:main:505 - Eval loss at step 3000: 3.7034199787663895, Eval perplexity: 40.585870093010406 + Update steps: 30%|██████▎ | 3001/10000 [05:47<10:35:05, 5.44s/it] Update steps: 30%|██████▌ | 3003/10000 [05:47<7:26:54, 3.83s/it] Update steps: 30%|██████▌ | 3005/10000 [05:47<5:15:12, 2.70s/it] Update steps: 30%|██████▌ | 3007/10000 [05:47<3:43:07, 1.91s/it] Update steps: 30%|██████▌ | 3009/10000 [05:47<2:38:40, 1.36s/it] Update steps: 30%|██████▌ | 3011/10000 [05:48<1:53:32, 1.03it/s] Update steps: 30%|██████▋ | 3013/10000 [05:48<1:21:55, 1.42it/s] Update steps: 30%|███████▏ | 3015/10000 [05:48<59:49, 1.95it/s] Update steps: 30%|███████▏ | 3017/10000 [05:48<44:20, 2.62it/s] Update steps: 30%|███████▏ | 3019/10000 [05:48<33:33, 3.47it/s] Update steps: 30%|███████▎ | 3021/10000 [05:48<25:59, 4.47it/s] Update steps: 30%|███████▎ | 3023/10000 [05:48<20:42, 5.61it/s] Update steps: 30%|███████▎ | 3025/10000 [05:49<17:04, 6.81it/s] Update steps: 30%|███████▎ | 3027/10000 [05:49<14:24, 8.06it/s] Update steps: 30%|███████▎ | 3029/10000 [05:49<12:40, 9.16it/s] Update steps: 30%|███████▎ | 3031/10000 [05:49<11:28, 10.12it/s] Update steps: 30%|███████▎ | 3033/10000 [05:49<10:36, 10.95it/s] Update steps: 30%|███████▎ | 3035/10000 [05:49<09:58, 11.63it/s] Update steps: 30%|███████▎ | 3037/10000 [05:49<09:33, 12.15it/s] Update steps: 30%|███████▎ | 3039/10000 [05:50<09:16, 12.51it/s] Update steps: 30%|███████▎ | 3041/10000 [05:50<08:59, 12.91it/s] Update steps: 30%|███████▎ | 3043/10000 [05:50<08:47, 13.18it/s] Update steps: 30%|███████▎ | 3045/10000 [05:50<08:41, 13.35it/s] Update steps: 30%|███████▎ | 3047/10000 [05:50<08:35, 13.49it/s] Update steps: 30%|███████▎ | 3049/10000 [05:50<08:34, 13.50it/s] Update steps: 31%|███████▎ | 3051/10000 [05:50<08:36, 13.45it/s] Update steps: 31%|███████▎ | 3053/10000 [05:51<08:35, 13.47it/s] Update steps: 31%|███████▎ | 3055/10000 [05:51<08:29, 13.63it/s] Update steps: 31%|███████▎ | 3057/10000 [05:51<08:29, 13.63it/s] Update steps: 31%|███████▎ | 3059/10000 [05:51<08:26, 13.72it/s] Update steps: 31%|███████▎ | 3061/10000 [05:51<08:25, 13.72it/s] Update steps: 31%|███████▎ | 3063/10000 [05:51<08:22, 13.81it/s] Update steps: 31%|███████▎ | 3065/10000 [05:51<08:20, 13.85it/s] Update steps: 31%|███████▎ | 3067/10000 [05:52<08:22, 13.80it/s] Update steps: 31%|███████▎ | 3069/10000 [05:52<08:20, 13.86it/s] Update steps: 31%|███████▎ | 3071/10000 [05:52<08:17, 13.91it/s] Update steps: 31%|███████▍ | 3073/10000 [05:52<08:19, 13.85it/s] Update steps: 31%|███████▍ | 3075/10000 [05:52<08:23, 13.74it/s] Update steps: 31%|███████▍ | 3077/10000 [05:52<08:26, 13.68it/s] Update steps: 31%|███████▍ | 3079/10000 [05:52<08:22, 13.77it/s] Update steps: 31%|███████▍ | 3081/10000 [05:53<08:19, 13.86it/s] Update steps: 31%|███████▍ | 3083/10000 [05:53<08:17, 13.90it/s] Update steps: 31%|███████▍ | 3085/10000 [05:53<08:20, 13.80it/s] Update steps: 31%|███████▍ | 3087/10000 [05:53<08:25, 13.69it/s] Update steps: 31%|███████▍ | 3089/10000 [05:53<08:23, 13.72it/s] Update steps: 31%|███████▍ | 3091/10000 [05:53<08:25, 13.67it/s] Update steps: 31%|███████▍ | 3093/10000 [05:54<08:26, 13.63it/s] Update steps: 31%|███████▍ | 3095/10000 [05:54<08:27, 13.61it/s] Update steps: 31%|███████▍ | 3097/10000 [05:54<08:24, 13.68it/s] Update steps: 31%|███████▍ | 3099/10000 [05:54<08:30, 13.53it/s] Update steps: 31%|███████▍ | 3101/10000 [05:54<08:26, 13.62it/s] Update steps: 31%|███████▍ | 3103/10000 [05:54<08:29, 13.53it/s] Update steps: 31%|███████▍ | 3105/10000 [05:54<08:27, 13.59it/s] Update steps: 31%|███████▍ | 3107/10000 [05:55<08:25, 13.65it/s] Update steps: 31%|███████▍ | 3109/10000 [05:55<08:22, 13.72it/s] Update steps: 31%|███████▍ | 3111/10000 [05:55<08:20, 13.77it/s] Update steps: 31%|███████▍ | 3113/10000 [05:55<08:17, 13.84it/s] Update steps: 31%|███████▍ | 3115/10000 [05:55<08:16, 13.86it/s] Update steps: 31%|███████▍ | 3117/10000 [05:55<08:19, 13.79it/s] Update steps: 31%|███████▍ | 3119/10000 [05:55<08:19, 13.77it/s] Update steps: 31%|███████▍ | 3121/10000 [05:56<08:21, 13.72it/s] Update steps: 31%|███████▍ | 3123/10000 [05:56<08:23, 13.66it/s] Update steps: 31%|███████▌ | 3125/10000 [05:56<08:25, 13.61it/s] Update steps: 31%|███████▌ | 3127/10000 [05:56<08:24, 13.62it/s] Update steps: 31%|███████▌ | 3129/10000 [05:56<08:24, 13.61it/s] Update steps: 31%|███████▌ | 3131/10000 [05:56<08:25, 13.59it/s] Update steps: 31%|███████▌ | 3133/10000 [05:56<08:20, 13.72it/s] Update steps: 31%|███████▌ | 3135/10000 [05:57<08:20, 13.72it/s] Update steps: 31%|███████▌ | 3137/10000 [05:57<08:22, 13.67it/s] Update steps: 31%|███████▌ | 3139/10000 [05:57<08:27, 13.52it/s] Update steps: 31%|███████▌ | 3141/10000 [05:57<08:55, 12.80it/s] Update steps: 31%|███████▌ | 3143/10000 [05:57<08:47, 12.99it/s] Update steps: 31%|███████▌ | 3145/10000 [05:57<08:38, 13.21it/s] Update steps: 31%|███████▌ | 3147/10000 [05:57<08:45, 13.04it/s] Update steps: 31%|███████▌ | 3149/10000 [05:58<08:56, 12.78it/s] Update steps: 32%|███████▌ | 3151/10000 [05:58<08:42, 13.10it/s] Update steps: 32%|███████▌ | 3153/10000 [05:58<09:09, 12.47it/s] Update steps: 32%|███████▌ | 3155/10000 [05:58<08:57, 12.74it/s] Update steps: 32%|███████▌ | 3157/10000 [05:58<09:00, 12.66it/s] Update steps: 32%|███████▌ | 3159/10000 [05:58<08:51, 12.86it/s] Update steps: 32%|███████▌ | 3161/10000 [05:59<08:59, 12.68it/s] Update steps: 32%|███████▌ | 3163/10000 [05:59<08:44, 13.04it/s] Update steps: 32%|███████▌ | 3165/10000 [05:59<08:59, 12.66it/s] Update steps: 32%|███████▌ | 3167/10000 [05:59<09:08, 12.46it/s] Update steps: 32%|███████▌ | 3169/10000 [05:59<09:02, 12.60it/s] Update steps: 32%|███████▌ | 3171/10000 [05:59<09:00, 12.64it/s] Update steps: 32%|███████▌ | 3173/10000 [06:00<08:51, 12.85it/s] Update steps: 32%|███████▌ | 3175/10000 [06:00<08:48, 12.91it/s] Update steps: 32%|███████▌ | 3177/10000 [06:00<08:53, 12.78it/s] Update steps: 32%|███████▋ | 3179/10000 [06:00<08:54, 12.75it/s] Update steps: 32%|███████▋ | 3181/10000 [06:00<08:54, 12.75it/s] Update steps: 32%|███████▋ | 3183/10000 [06:00<08:44, 13.00it/s] Update steps: 32%|███████▋ | 3185/10000 [06:00<08:51, 12.81it/s] Update steps: 32%|███████▋ | 3187/10000 [06:01<08:57, 12.68it/s] Update steps: 32%|███████▋ | 3189/10000 [06:01<08:52, 12.78it/s] Update steps: 32%|███████▋ | 3191/10000 [06:01<08:47, 12.91it/s] Update steps: 32%|███████▋ | 3193/10000 [06:01<08:44, 12.97it/s] Update steps: 32%|███████▋ | 3195/10000 [06:01<08:39, 13.09it/s] Update steps: 32%|███████▋ | 3197/10000 [06:01<08:38, 13.12it/s] Update steps: 32%|███████▋ | 3199/10000 [06:02<08:34, 13.22it/s] Update steps: 32%|███████▋ | 3201/10000 [06:02<08:35, 13.20it/s] Update steps: 32%|███████▋ | 3203/10000 [06:02<08:35, 13.19it/s] Update steps: 32%|███████▋ | 3205/10000 [06:02<08:33, 13.22it/s] Update steps: 32%|███████▋ | 3207/10000 [06:02<08:31, 13.28it/s] Update steps: 32%|███████▋ | 3209/10000 [06:02<08:41, 13.01it/s] Update steps: 32%|███████▋ | 3211/10000 [06:02<08:48, 12.85it/s] Update steps: 32%|███████▋ | 3213/10000 [06:03<08:41, 13.02it/s] Update steps: 32%|███████▋ | 3215/10000 [06:03<08:37, 13.12it/s] Update steps: 32%|███████▋ | 3217/10000 [06:03<08:46, 12.88it/s] Update steps: 32%|███████▋ | 3219/10000 [06:03<08:41, 13.01it/s] Update steps: 32%|███████▋ | 3221/10000 [06:03<09:02, 12.50it/s] Update steps: 32%|███████▋ | 3223/10000 [06:03<09:08, 12.35it/s] Update steps: 32%|███████▋ | 3225/10000 [06:04<08:55, 12.64it/s] Update steps: 32%|███████▋ | 3227/10000 [06:04<08:46, 12.85it/s] Update steps: 32%|███████▋ | 3229/10000 [06:04<08:44, 12.92it/s] Update steps: 32%|███████▊ | 3231/10000 [06:04<08:39, 13.04it/s] Update steps: 32%|███████▊ | 3233/10000 [06:04<08:46, 12.86it/s] Update steps: 32%|███████▊ | 3235/10000 [06:04<08:48, 12.79it/s] Update steps: 32%|███████▊ | 3237/10000 [06:04<08:38, 13.05it/s] Update steps: 32%|███████▊ | 3239/10000 [06:05<08:34, 13.14it/s] Update steps: 32%|███████▊ | 3241/10000 [06:05<08:42, 12.94it/s] Update steps: 32%|███████▊ | 3243/10000 [06:05<08:28, 13.28it/s] Update steps: 32%|███████▊ | 3245/10000 [06:05<08:37, 13.05it/s] Update steps: 32%|███████▊ | 3247/10000 [06:05<08:37, 13.06it/s] Update steps: 32%|███████▊ | 3249/10000 [06:05<08:43, 12.90it/s] Update steps: 33%|███████▊ | 3251/10000 [06:06<09:05, 12.37it/s] Update steps: 33%|███████▊ | 3253/10000 [06:06<08:48, 12.78it/s] Update steps: 33%|███████▊ | 3255/10000 [06:06<08:41, 12.93it/s] Update steps: 33%|███████▊ | 3257/10000 [06:06<08:40, 12.95it/s] Update steps: 33%|███████▊ | 3259/10000 [06:06<08:46, 12.81it/s] Update steps: 33%|███████▊ | 3261/10000 [06:06<08:42, 12.89it/s] Update steps: 33%|███████▊ | 3263/10000 [06:07<08:44, 12.85it/s] Update steps: 33%|███████▊ | 3265/10000 [06:07<08:55, 12.57it/s] Update steps: 33%|███████▊ | 3267/10000 [06:07<08:55, 12.58it/s] Update steps: 33%|███████▊ | 3269/10000 [06:07<08:47, 12.76it/s] Update steps: 33%|███████▊ | 3271/10000 [06:07<08:41, 12.91it/s] Update steps: 33%|███████▊ | 3273/10000 [06:07<08:38, 12.98it/s] Update steps: 33%|███████▊ | 3275/10000 [06:07<08:41, 12.90it/s] Update steps: 33%|███████▊ | 3277/10000 [06:08<08:35, 13.03it/s] Update steps: 33%|███████▊ | 3279/10000 [06:08<08:33, 13.09it/s] Update steps: 33%|███████▊ | 3281/10000 [06:08<08:40, 12.90it/s] Update steps: 33%|███████▉ | 3283/10000 [06:08<08:51, 12.64it/s] Update steps: 33%|███████▉ | 3285/10000 [06:08<08:39, 12.92it/s] Update steps: 33%|███████▉ | 3287/10000 [06:08<08:57, 12.50it/s] Update steps: 33%|███████▉ | 3289/10000 [06:09<08:59, 12.44it/s] Update steps: 33%|███████▉ | 3291/10000 [06:09<08:49, 12.66it/s] Update steps: 33%|███████▉ | 3293/10000 [06:09<09:24, 11.89it/s] Update steps: 33%|███████▉ | 3295/10000 [06:09<09:01, 12.38it/s] Update steps: 33%|███████▉ | 3297/10000 [06:09<09:06, 12.26it/s] Update steps: 33%|███████▉ | 3299/10000 [06:09<09:02, 12.34it/s] Update steps: 33%|███████▉ | 3301/10000 [06:10<08:46, 12.72it/s] Update steps: 33%|███████▉ | 3303/10000 [06:10<09:04, 12.29it/s] Update steps: 33%|███████▉ | 3305/10000 [06:10<08:51, 12.60it/s] Update steps: 33%|███████▉ | 3307/10000 [06:10<08:41, 12.83it/s] Update steps: 33%|███████▉ | 3309/10000 [06:10<08:51, 12.58it/s] Update steps: 33%|███████▉ | 3311/10000 [06:10<08:40, 12.85it/s] Update steps: 33%|███████▉ | 3313/10000 [06:10<08:31, 13.07it/s] Update steps: 33%|███████▉ | 3315/10000 [06:11<08:34, 12.99it/s] Update steps: 33%|███████▉ | 3317/10000 [06:11<08:31, 13.08it/s] Update steps: 33%|███████▉ | 3319/10000 [06:11<08:21, 13.33it/s] Update steps: 33%|███████▉ | 3321/10000 [06:11<08:36, 12.94it/s] Update steps: 33%|███████▉ | 3323/10000 [06:11<08:35, 12.96it/s] Update steps: 33%|███████▉ | 3325/10000 [06:11<08:24, 13.23it/s] Update steps: 33%|███████▉ | 3327/10000 [06:12<08:27, 13.15it/s] Update steps: 33%|███████▉ | 3329/10000 [06:12<08:52, 12.54it/s] Update steps: 33%|███████▉ | 3331/10000 [06:12<08:56, 12.43it/s] Update steps: 33%|███████▉ | 3333/10000 [06:12<08:41, 12.77it/s] Update steps: 33%|████████ | 3335/10000 [06:12<08:35, 12.92it/s] Update steps: 33%|████████ | 3337/10000 [06:12<08:47, 12.62it/s] Update steps: 33%|████████ | 3339/10000 [06:12<08:50, 12.56it/s] Update steps: 33%|████████ | 3341/10000 [06:13<08:38, 12.85it/s] Update steps: 33%|████████ | 3343/10000 [06:13<08:40, 12.79it/s] Update steps: 33%|████████ | 3345/10000 [06:13<08:48, 12.59it/s] Update steps: 33%|████████ | 3347/10000 [06:13<08:42, 12.72it/s] Update steps: 33%|████████ | 3349/10000 [06:13<08:32, 12.98it/s] Update steps: 34%|████████ | 3351/10000 [06:13<08:25, 13.16it/s] Update steps: 34%|████████ | 3353/10000 [06:14<08:49, 12.56it/s] Update steps: 34%|████████ | 3355/10000 [06:14<08:54, 12.44it/s] Update steps: 34%|████████ | 3357/10000 [06:14<08:36, 12.86it/s] Update steps: 34%|████████ | 3359/10000 [06:14<08:30, 13.00it/s] Update steps: 34%|████████ | 3361/10000 [06:14<08:29, 13.04it/s] Update steps: 34%|████████ | 3363/10000 [06:14<08:33, 12.93it/s] Update steps: 34%|████████ | 3365/10000 [06:15<08:44, 12.65it/s] Update steps: 34%|████████ | 3367/10000 [06:15<08:28, 13.04it/s] Update steps: 34%|████████ | 3369/10000 [06:15<08:22, 13.19it/s] Update steps: 34%|████████ | 3371/10000 [06:15<08:16, 13.35it/s] Update steps: 34%|████████ | 3373/10000 [06:15<08:45, 12.60it/s] Update steps: 34%|████████ | 3375/10000 [06:15<08:44, 12.63it/s] Update steps: 34%|████████ | 3377/10000 [06:15<08:41, 12.69it/s] Update steps: 34%|████████ | 3379/10000 [06:16<08:28, 13.02it/s] Update steps: 34%|████████ | 3381/10000 [06:16<08:17, 13.30it/s] Update steps: 34%|████████ | 3383/10000 [06:16<08:12, 13.45it/s] Update steps: 34%|████████ | 3385/10000 [06:16<08:26, 13.06it/s] Update steps: 34%|████████▏ | 3387/10000 [06:16<08:28, 13.00it/s] Update steps: 34%|████████▏ | 3389/10000 [06:16<08:46, 12.55it/s] Update steps: 34%|████████▏ | 3391/10000 [06:17<08:42, 12.65it/s] Update steps: 34%|████████▏ | 3393/10000 [06:17<08:26, 13.04it/s] Update steps: 34%|████████▏ | 3395/10000 [06:17<08:15, 13.33it/s] Update steps: 34%|████████▏ | 3397/10000 [06:17<08:08, 13.50it/s] Update steps: 34%|████████▏ | 3399/10000 [06:17<08:25, 13.05it/s] Update steps: 34%|████████▏ | 3401/10000 [06:17<08:35, 12.80it/s] Update steps: 34%|████████▏ | 3403/10000 [06:17<08:37, 12.76it/s] Update steps: 34%|████████▏ | 3405/10000 [06:18<08:31, 12.90it/s] Update steps: 34%|████████▏ | 3407/10000 [06:18<08:25, 13.03it/s] Update steps: 34%|████████▏ | 3409/10000 [06:18<08:14, 13.33it/s] Update steps: 34%|████████▏ | 3411/10000 [06:18<08:11, 13.41it/s] Update steps: 34%|████████▏ | 3413/10000 [06:18<08:05, 13.57it/s] Update steps: 34%|████████▏ | 3415/10000 [06:18<08:19, 13.17it/s] Update steps: 34%|████████▏ | 3417/10000 [06:19<08:36, 12.73it/s] Update steps: 34%|████████▏ | 3419/10000 [06:19<08:43, 12.57it/s] Update steps: 34%|████████▏ | 3421/10000 [06:19<08:47, 12.47it/s] Update steps: 34%|████████▏ | 3423/10000 [06:19<08:40, 12.64it/s] Update steps: 34%|████████▏ | 3425/10000 [06:19<08:24, 13.03it/s] Update steps: 34%|████████▏ | 3427/10000 [06:19<08:13, 13.31it/s] Update steps: 34%|████████▏ | 3429/10000 [06:19<08:10, 13.40it/s] Update steps: 34%|████████▏ | 3431/10000 [06:20<08:08, 13.43it/s] Update steps: 34%|████████▏ | 3433/10000 [06:20<08:23, 13.04it/s] Update steps: 34%|████████▏ | 3435/10000 [06:20<08:43, 12.55it/s] Update steps: 34%|████████▏ | 3437/10000 [06:20<08:47, 12.45it/s] Update steps: 34%|████████▎ | 3439/10000 [06:20<08:42, 12.56it/s] Update steps: 34%|████████▎ | 3441/10000 [06:20<08:46, 12.46it/s] Update steps: 34%|████████▎ | 3443/10000 [06:21<08:58, 12.17it/s] Update steps: 34%|████████▎ | 3445/10000 [06:21<08:40, 12.59it/s] Update steps: 34%|████████▎ | 3447/10000 [06:21<08:25, 12.96it/s] Update steps: 34%|████████▎ | 3449/10000 [06:21<08:15, 13.23it/s] Update steps: 35%|████████▎ | 3451/10000 [06:21<08:11, 13.34it/s] Update steps: 35%|████████▎ | 3453/10000 [06:21<08:03, 13.55it/s] Update steps: 35%|████████▎ | 3455/10000 [06:21<08:01, 13.59it/s] Update steps: 35%|████████▎ | 3457/10000 [06:22<08:14, 13.23it/s] Update steps: 35%|████████▎ | 3459/10000 [06:22<08:20, 13.06it/s] Update steps: 35%|████████▎ | 3461/10000 [06:22<08:25, 12.95it/s] Update steps: 35%|████████▎ | 3463/10000 [06:22<08:42, 12.51it/s] Update steps: 35%|████████▎ | 3465/10000 [06:22<08:51, 12.28it/s] Update steps: 35%|████████▎ | 3467/10000 [06:22<08:49, 12.33it/s] Update steps: 35%|████████▎ | 3469/10000 [06:23<08:53, 12.25it/s] Update steps: 35%|████████▎ | 3471/10000 [06:23<08:38, 12.59it/s] Update steps: 35%|████████▎ | 3473/10000 [06:23<08:23, 12.97it/s] Update steps: 35%|████████▎ | 3475/10000 [06:23<08:11, 13.26it/s] Update steps: 35%|████████▎ | 3477/10000 [06:23<08:06, 13.41it/s] Update steps: 35%|████████▎ | 3479/10000 [06:23<08:07, 13.37it/s] Update steps: 35%|████████▎ | 3481/10000 [06:23<08:04, 13.47it/s] Update steps: 35%|████████▎ | 3483/10000 [06:24<08:02, 13.51it/s] Update steps: 35%|████████▎ | 3485/10000 [06:24<08:00, 13.56it/s] Update steps: 35%|████████▎ | 3487/10000 [06:24<08:06, 13.38it/s] Update steps: 35%|████████▎ | 3489/10000 [06:24<08:22, 12.95it/s] Update steps: 35%|████████▍ | 3491/10000 [06:24<08:24, 12.91it/s] Update steps: 35%|████████▍ | 3493/10000 [06:24<08:42, 12.46it/s] Update steps: 35%|████████▍ | 3495/10000 [06:25<08:48, 12.32it/s] Update steps: 35%|████████▍ | 3497/10000 [06:25<08:51, 12.23it/s] Update steps: 35%|████████▍ | 3499/10000 [06:25<08:48, 12.30it/s] Update steps: 35%|████████▍ | 3501/10000 [06:25<08:51, 12.23it/s] Update steps: 35%|████████▍ | 3503/10000 [06:25<08:48, 12.28it/s] Update steps: 35%|████████▍ | 3505/10000 [06:25<08:44, 12.39it/s] Update steps: 35%|████████▍ | 3507/10000 [06:26<08:34, 12.61it/s] Update steps: 35%|████████▍ | 3509/10000 [06:26<08:19, 13.00it/s] Update steps: 35%|████████▍ | 3511/10000 [06:26<08:08, 13.27it/s] Update steps: 35%|████████▍ | 3513/10000 [06:26<08:05, 13.37it/s] Update steps: 35%|████████▍ | 3515/10000 [06:26<07:58, 13.54it/s] Update steps: 35%|████████▍ | 3517/10000 [06:26<07:59, 13.52it/s] Update steps: 35%|████████▍ | 3519/10000 [06:26<07:54, 13.66it/s] Update steps: 35%|████████▍ | 3521/10000 [06:27<07:59, 13.52it/s] Update steps: 35%|████████▍ | 3523/10000 [06:27<07:54, 13.65it/s] Update steps: 35%|████████▍ | 3525/10000 [06:27<07:55, 13.61it/s] Update steps: 35%|████████▍ | 3527/10000 [06:27<07:51, 13.73it/s] Update steps: 35%|████████▍ | 3529/10000 [06:27<07:53, 13.66it/s] Update steps: 35%|████████▍ | 3531/10000 [06:27<07:53, 13.65it/s] Update steps: 35%|████████▍ | 3533/10000 [06:27<08:16, 13.03it/s] Update steps: 35%|████████▍ | 3535/10000 [06:28<08:19, 12.95it/s] Update steps: 35%|████████▍ | 3537/10000 [06:28<08:38, 12.47it/s] Update steps: 35%|████████▍ | 3539/10000 [06:28<08:44, 12.33it/s] Update steps: 35%|████████▍ | 3541/10000 [06:28<08:49, 12.21it/s] Update steps: 35%|████████▌ | 3543/10000 [06:28<08:51, 12.15it/s] Update steps: 35%|████████▌ | 3545/10000 [06:28<08:43, 12.33it/s] Update steps: 35%|████████▌ | 3547/10000 [06:29<08:39, 12.43it/s] Update steps: 35%|████████▌ | 3549/10000 [06:29<08:42, 12.34it/s] Update steps: 36%|████████▌ | 3551/10000 [06:29<08:46, 12.24it/s] Update steps: 36%|████████▌ | 3553/10000 [06:29<08:39, 12.41it/s] Update steps: 36%|████████▌ | 3555/10000 [06:29<08:43, 12.32it/s] Update steps: 36%|████████▌ | 3557/10000 [06:29<08:45, 12.25it/s] Update steps: 36%|████████▌ | 3559/10000 [06:30<08:50, 12.14it/s] Update steps: 36%|████████▌ | 3561/10000 [06:30<09:05, 11.79it/s] Update steps: 36%|████████▌ | 3563/10000 [06:30<09:05, 11.81it/s] Update steps: 36%|████████▌ | 3565/10000 [06:30<08:58, 11.96it/s] Update steps: 36%|████████▌ | 3567/10000 [06:30<08:37, 12.43it/s] Update steps: 36%|████████▌ | 3569/10000 [06:30<08:26, 12.70it/s] Update steps: 36%|████████▌ | 3571/10000 [06:31<08:14, 13.00it/s] Update steps: 36%|████████▌ | 3573/10000 [06:31<08:03, 13.28it/s] Update steps: 36%|████████▌ | 3575/10000 [06:31<07:55, 13.50it/s] Update steps: 36%|████████▌ | 3577/10000 [06:31<07:54, 13.55it/s] Update steps: 36%|████████▌ | 3579/10000 [06:31<07:52, 13.58it/s] Update steps: 36%|████████▌ | 3581/10000 [06:31<07:48, 13.70it/s] Update steps: 36%|████████▌ | 3583/10000 [06:31<07:49, 13.65it/s] Update steps: 36%|████████▌ | 3585/10000 [06:32<07:49, 13.67it/s] Update steps: 36%|████████▌ | 3587/10000 [06:32<07:47, 13.71it/s] Update steps: 36%|████████▌ | 3589/10000 [06:32<07:49, 13.65it/s] Update steps: 36%|████████▌ | 3591/10000 [06:32<07:46, 13.73it/s] Update steps: 36%|████████▌ | 3593/10000 [06:32<07:49, 13.66it/s] Update steps: 36%|████████▋ | 3595/10000 [06:32<07:48, 13.67it/s] Update steps: 36%|████████▋ | 3597/10000 [06:32<07:46, 13.72it/s] Update steps: 36%|████████▋ | 3599/10000 [06:33<07:48, 13.65it/s] Update steps: 36%|████████▋ | 3601/10000 [06:33<07:51, 13.56it/s] Update steps: 36%|████████▋ | 3603/10000 [06:33<07:51, 13.55it/s] Update steps: 36%|████████▋ | 3605/10000 [06:33<07:54, 13.48it/s] Update steps: 36%|████████▋ | 3607/10000 [06:33<07:52, 13.53it/s] Update steps: 36%|████████▋ | 3609/10000 [06:33<07:53, 13.50it/s] Update steps: 36%|████████▋ | 3611/10000 [06:33<07:51, 13.54it/s] Update steps: 36%|████████▋ | 3613/10000 [06:34<07:46, 13.70it/s] Update steps: 36%|████████▋ | 3615/10000 [06:34<07:43, 13.77it/s] Update steps: 36%|████████▋ | 3617/10000 [06:34<08:08, 13.08it/s] Update steps: 36%|████████▋ | 3619/10000 [06:34<08:12, 12.96it/s] Update steps: 36%|████████▋ | 3621/10000 [06:34<08:25, 12.63it/s] Update steps: 36%|████████▋ | 3623/10000 [06:34<08:22, 12.70it/s] Update steps: 36%|████████▋ | 3625/10000 [06:35<08:30, 12.49it/s] Update steps: 36%|████████▋ | 3627/10000 [06:35<08:26, 12.58it/s] Update steps: 36%|████████▋ | 3629/10000 [06:35<08:39, 12.27it/s] Update steps: 36%|████████▋ | 3631/10000 [06:35<08:39, 12.25it/s] Update steps: 36%|████████▋ | 3633/10000 [06:35<08:36, 12.32it/s] Update steps: 36%|████████▋ | 3635/10000 [06:35<08:31, 12.45it/s] Update steps: 36%|████████▋ | 3637/10000 [06:36<08:30, 12.46it/s] Update steps: 36%|████████▋ | 3639/10000 [06:36<08:33, 12.40it/s] Update steps: 36%|████████▋ | 3641/10000 [06:36<08:34, 12.36it/s] Update steps: 36%|████████▋ | 3643/10000 [06:36<08:34, 12.34it/s] Update steps: 36%|████████▋ | 3645/10000 [06:36<08:32, 12.40it/s] Update steps: 36%|████████▊ | 3647/10000 [06:36<08:34, 12.34it/s] Update steps: 36%|████████▊ | 3649/10000 [06:37<08:45, 12.09it/s] Update steps: 37%|████████▊ | 3651/10000 [06:37<08:36, 12.30it/s] Update steps: 37%|████████▊ | 3653/10000 [06:37<08:48, 12.01it/s] Update steps: 37%|████████▊ | 3655/10000 [06:37<08:39, 12.22it/s] Update steps: 37%|████████▊ | 3657/10000 [06:37<08:30, 12.42it/s] Update steps: 37%|████████▊ | 3659/10000 [06:37<08:34, 12.33it/s] Update steps: 37%|████████▊ | 3661/10000 [06:37<08:38, 12.24it/s] Update steps: 37%|████████▊ | 3663/10000 [06:38<08:41, 12.14it/s] Update steps: 37%|████████▊ | 3665/10000 [06:38<08:47, 12.00it/s] Update steps: 37%|████████▊ | 3667/10000 [06:38<08:53, 11.87it/s] Update steps: 37%|████████▊ | 3669/10000 [06:38<08:54, 11.84it/s] Update steps: 37%|████████▊ | 3671/10000 [06:38<08:50, 11.93it/s] Update steps: 37%|████████▊ | 3673/10000 [06:39<08:53, 11.86it/s] Update steps: 37%|████████▊ | 3675/10000 [06:39<08:40, 12.14it/s] Update steps: 37%|████████▊ | 3677/10000 [06:39<08:33, 12.31it/s] Update steps: 37%|████████▊ | 3679/10000 [06:39<08:45, 12.03it/s] Update steps: 37%|████████▊ | 3681/10000 [06:39<08:43, 12.07it/s] Update steps: 37%|████████▊ | 3683/10000 [06:39<08:41, 12.12it/s] Update steps: 37%|████████▊ | 3685/10000 [06:39<08:38, 12.18it/s] Update steps: 37%|████████▊ | 3687/10000 [06:40<08:39, 12.16it/s] Update steps: 37%|████████▊ | 3689/10000 [06:40<08:30, 12.36it/s] Update steps: 37%|████████▊ | 3691/10000 [06:40<08:24, 12.51it/s] Update steps: 37%|████████▊ | 3693/10000 [06:40<08:19, 12.63it/s] Update steps: 37%|████████▊ | 3695/10000 [06:40<08:18, 12.65it/s] Update steps: 37%|████████▊ | 3697/10000 [06:40<08:27, 12.42it/s] Update steps: 37%|████████▉ | 3699/10000 [06:41<08:25, 12.46it/s] Update steps: 37%|████████▉ | 3701/10000 [06:41<08:35, 12.22it/s] Update steps: 37%|████████▉ | 3703/10000 [06:41<08:42, 12.05it/s] Update steps: 37%|████████▉ | 3705/10000 [06:41<08:49, 11.89it/s] Update steps: 37%|████████▉ | 3707/10000 [06:41<08:54, 11.78it/s] Update steps: 37%|████████▉ | 3709/10000 [06:41<08:55, 11.75it/s] Update steps: 37%|████████▉ | 3711/10000 [06:42<08:43, 12.02it/s] Update steps: 37%|████████▉ | 3713/10000 [06:42<08:35, 12.20it/s] Update steps: 37%|████████▉ | 3715/10000 [06:42<08:37, 12.14it/s] Update steps: 37%|████████▉ | 3717/10000 [06:42<08:34, 12.21it/s] Update steps: 37%|████████▉ | 3719/10000 [06:42<08:42, 12.02it/s] Update steps: 37%|████████▉ | 3721/10000 [06:42<08:38, 12.10it/s] Update steps: 37%|████████▉ | 3723/10000 [06:43<08:30, 12.30it/s] Update steps: 37%|████████▉ | 3725/10000 [06:43<08:25, 12.41it/s] Update steps: 37%|████████▉ | 3727/10000 [06:43<08:06, 12.89it/s] Update steps: 37%|████████▉ | 3729/10000 [06:43<07:54, 13.23it/s] Update steps: 37%|████████▉ | 3731/10000 [06:43<07:45, 13.46it/s] Update steps: 37%|████████▉ | 3733/10000 [06:43<07:43, 13.53it/s] Update steps: 37%|████████▉ | 3735/10000 [06:43<07:40, 13.61it/s] Update steps: 37%|████████▉ | 3737/10000 [06:44<07:35, 13.76it/s] Update steps: 37%|████████▉ | 3739/10000 [06:44<07:33, 13.80it/s] Update steps: 37%|████████▉ | 3741/10000 [06:44<07:31, 13.87it/s] Update steps: 37%|████████▉ | 3743/10000 [06:44<07:33, 13.80it/s] Update steps: 37%|████████▉ | 3745/10000 [06:44<07:32, 13.81it/s] Update steps: 37%|████████▉ | 3747/10000 [06:44<07:35, 13.74it/s] Update steps: 37%|████████▉ | 3749/10000 [06:44<07:31, 13.83it/s] Update steps: 38%|█████████ | 3751/10000 [06:45<07:29, 13.89it/s] Update steps: 38%|█████████ | 3753/10000 [06:45<07:32, 13.80it/s] Update steps: 38%|█████████ | 3755/10000 [06:45<07:31, 13.84it/s] Update steps: 38%|█████████ | 3757/10000 [06:45<07:33, 13.76it/s] Update steps: 38%|█████████ | 3759/10000 [06:45<07:34, 13.73it/s] Update steps: 38%|█████████ | 3761/10000 [06:45<07:34, 13.73it/s] Update steps: 38%|█████████ | 3763/10000 [06:45<07:31, 13.81it/s] Update steps: 38%|█████████ | 3765/10000 [06:46<07:35, 13.69it/s] Update steps: 38%|█████████ | 3767/10000 [06:46<07:34, 13.73it/s] Update steps: 38%|█████████ | 3769/10000 [06:46<07:33, 13.75it/s] Update steps: 38%|█████████ | 3771/10000 [06:46<07:29, 13.86it/s] Update steps: 38%|█████████ | 3773/10000 [06:46<07:28, 13.88it/s] Update steps: 38%|█████████ | 3775/10000 [06:46<07:31, 13.78it/s] Update steps: 38%|█████████ | 3777/10000 [06:47<07:31, 13.77it/s] Update steps: 38%|█████████ | 3779/10000 [06:47<07:32, 13.76it/s] Update steps: 38%|█████████ | 3781/10000 [06:47<07:34, 13.67it/s] Update steps: 38%|█████████ | 3783/10000 [06:47<07:39, 13.54it/s] Update steps: 38%|█████████ | 3785/10000 [06:47<07:42, 13.45it/s] Update steps: 38%|█████████ | 3787/10000 [06:47<07:37, 13.59it/s] Update steps: 38%|█████████ | 3789/10000 [06:47<07:31, 13.74it/s] Update steps: 38%|█████████ | 3791/10000 [06:48<07:32, 13.73it/s] Update steps: 38%|█████████ | 3793/10000 [06:48<07:34, 13.65it/s] Update steps: 38%|█████████ | 3795/10000 [06:48<07:31, 13.75it/s] Update steps: 38%|█████████ | 3797/10000 [06:48<07:30, 13.78it/s] Update steps: 38%|█████████ | 3799/10000 [06:48<07:30, 13.78it/s] Update steps: 38%|█████████ | 3801/10000 [06:48<07:30, 13.77it/s] Update steps: 38%|█████████▏ | 3803/10000 [06:48<07:30, 13.75it/s] Update steps: 38%|█████████▏ | 3805/10000 [06:49<07:28, 13.80it/s] Update steps: 38%|█████████▏ | 3807/10000 [06:49<07:33, 13.65it/s] Update steps: 38%|█████████▏ | 3809/10000 [06:49<07:35, 13.60it/s] Update steps: 38%|█████████▏ | 3811/10000 [06:49<07:34, 13.61it/s] Update steps: 38%|█████████▏ | 3813/10000 [06:49<07:32, 13.68it/s] Update steps: 38%|█████████▏ | 3815/10000 [06:49<07:36, 13.55it/s] Update steps: 38%|█████████▏ | 3817/10000 [06:49<07:36, 13.55it/s] Update steps: 38%|█████████▏ | 3819/10000 [06:50<07:34, 13.60it/s] Update steps: 38%|█████████▏ | 3821/10000 [06:50<07:32, 13.67it/s] Update steps: 38%|█████████▏ | 3823/10000 [06:50<07:32, 13.65it/s] Update steps: 38%|█████████▏ | 3825/10000 [06:50<07:31, 13.69it/s] Update steps: 38%|█████████▏ | 3827/10000 [06:50<07:27, 13.80it/s] Update steps: 38%|█████████▏ | 3829/10000 [06:50<07:26, 13.83it/s] Update steps: 38%|█████████▏ | 3831/10000 [06:50<07:31, 13.66it/s] Update steps: 38%|█████████▏ | 3833/10000 [06:51<07:35, 13.55it/s] Update steps: 38%|█████████▏ | 3835/10000 [06:51<07:36, 13.49it/s] Update steps: 38%|█████████▏ | 3837/10000 [06:51<07:34, 13.55it/s] Update steps: 38%|█████████▏ | 3839/10000 [06:51<07:30, 13.69it/s] Update steps: 38%|█████████▏ | 3841/10000 [06:51<07:26, 13.79it/s] Update steps: 38%|█████████▏ | 3843/10000 [06:51<07:26, 13.79it/s] Update steps: 38%|█████████▏ | 3845/10000 [06:51<07:28, 13.72it/s] Update steps: 38%|█████████▏ | 3847/10000 [06:52<07:31, 13.61it/s] Update steps: 38%|█████████▏ | 3849/10000 [06:52<07:31, 13.62it/s] Update steps: 39%|█████████▏ | 3851/10000 [06:52<07:27, 13.73it/s] Update steps: 39%|█████████▏ | 3853/10000 [06:52<07:27, 13.74it/s] Update steps: 39%|█████████▎ | 3855/10000 [06:52<07:25, 13.78it/s] Update steps: 39%|█████████▎ | 3857/10000 [06:52<07:31, 13.62it/s] Update steps: 39%|█████████▎ | 3859/10000 [06:53<07:27, 13.71it/s] Update steps: 39%|█████████▎ | 3861/10000 [06:53<07:28, 13.68it/s] Update steps: 39%|█████████▎ | 3863/10000 [06:53<07:30, 13.61it/s] Update steps: 39%|█████████▎ | 3865/10000 [06:53<07:32, 13.57it/s] Update steps: 39%|█████████▎ | 3867/10000 [06:53<07:32, 13.57it/s] Update steps: 39%|█████████▎ | 3869/10000 [06:53<07:31, 13.58it/s] Update steps: 39%|█████████▎ | 3871/10000 [06:53<07:33, 13.51it/s] Update steps: 39%|█████████▎ | 3873/10000 [06:54<07:35, 13.45it/s] Update steps: 39%|█████████▎ | 3875/10000 [06:54<07:31, 13.56it/s] Update steps: 39%|█████████▎ | 3877/10000 [06:54<07:27, 13.68it/s] Update steps: 39%|█████████▎ | 3879/10000 [06:54<07:24, 13.78it/s] Update steps: 39%|█████████▎ | 3881/10000 [06:54<07:22, 13.84it/s] Update steps: 39%|█████████▎ | 3883/10000 [06:54<07:22, 13.83it/s] Update steps: 39%|█████████▎ | 3885/10000 [06:54<07:24, 13.74it/s] Update steps: 39%|█████████▎ | 3887/10000 [06:55<07:26, 13.71it/s] Update steps: 39%|█████████▎ | 3889/10000 [06:55<07:28, 13.61it/s] Update steps: 39%|█████████▎ | 3891/10000 [06:55<07:29, 13.59it/s] Update steps: 39%|█████████▎ | 3893/10000 [06:55<07:26, 13.69it/s] Update steps: 39%|█████████▎ | 3895/10000 [06:55<07:27, 13.63it/s] Update steps: 39%|█████████▎ | 3897/10000 [06:55<07:24, 13.74it/s] Update steps: 39%|█████████▎ | 3899/10000 [06:55<07:27, 13.65it/s] Update steps: 39%|█████████▎ | 3901/10000 [06:56<07:24, 13.71it/s] Update steps: 39%|█████████▎ | 3903/10000 [06:56<07:22, 13.79it/s] Update steps: 39%|█████████▎ | 3905/10000 [06:56<07:22, 13.79it/s] Update steps: 39%|█████████▍ | 3907/10000 [06:56<07:24, 13.71it/s] Update steps: 39%|█████████▍ | 3909/10000 [06:56<07:23, 13.72it/s] Update steps: 39%|█████████▍ | 3911/10000 [06:56<07:23, 13.72it/s] Update steps: 39%|█████████▍ | 3913/10000 [06:56<07:28, 13.56it/s] Update steps: 39%|█████████▍ | 3915/10000 [06:57<07:34, 13.38it/s] Update steps: 39%|█████████▍ | 3917/10000 [06:57<07:33, 13.42it/s] Update steps: 39%|█████████▍ | 3919/10000 [06:57<07:31, 13.46it/s] Update steps: 39%|█████████▍ | 3921/10000 [06:57<07:30, 13.48it/s] Update steps: 39%|█████████▍ | 3923/10000 [06:57<07:27, 13.58it/s] Update steps: 39%|█████████▍ | 3925/10000 [06:57<07:27, 13.58it/s] Update steps: 39%|█████████▍ | 3927/10000 [06:57<07:28, 13.54it/s] Update steps: 39%|█████████▍ | 3929/10000 [06:58<07:28, 13.54it/s] Update steps: 39%|█████████▍ | 3931/10000 [06:58<07:26, 13.59it/s] Update steps: 39%|█████████▍ | 3933/10000 [06:58<07:22, 13.72it/s] Update steps: 39%|█████████▍ | 3935/10000 [06:58<07:19, 13.79it/s] Update steps: 39%|█████████▍ | 3937/10000 [06:58<07:20, 13.77it/s] Update steps: 39%|█████████▍ | 3939/10000 [06:58<07:26, 13.57it/s] Update steps: 39%|█████████▍ | 3941/10000 [06:59<07:24, 13.62it/s] Update steps: 39%|█████████▍ | 3943/10000 [06:59<07:24, 13.63it/s] Update steps: 39%|█████████▍ | 3945/10000 [06:59<07:27, 13.54it/s] Update steps: 39%|█████████▍ | 3947/10000 [06:59<07:23, 13.64it/s] Update steps: 39%|█████████▍ | 3949/10000 [06:59<07:23, 13.64it/s] Update steps: 40%|█████████▍ | 3951/10000 [06:59<07:19, 13.77it/s] Update steps: 40%|█████████▍ | 3953/10000 [06:59<07:18, 13.79it/s] Update steps: 40%|█████████▍ | 3955/10000 [07:00<07:18, 13.79it/s] Update steps: 40%|█████████▍ | 3957/10000 [07:00<07:20, 13.71it/s] Update steps: 40%|█████████▌ | 3959/10000 [07:00<07:21, 13.67it/s] Update steps: 40%|█████████▌ | 3961/10000 [07:00<07:19, 13.74it/s] Update steps: 40%|█████████▌ | 3963/10000 [07:00<07:23, 13.63it/s] Update steps: 40%|█████████▌ | 3965/10000 [07:00<07:22, 13.63it/s] Update steps: 40%|█████████▌ | 3967/10000 [07:00<07:23, 13.59it/s] Update steps: 40%|█████████▌ | 3969/10000 [07:01<07:25, 13.55it/s] Update steps: 40%|█████████▌ | 3971/10000 [07:01<07:20, 13.69it/s] Update steps: 40%|█████████▌ | 3973/10000 [07:01<07:22, 13.62it/s] Update steps: 40%|█████████▌ | 3975/10000 [07:01<07:20, 13.66it/s] Update steps: 40%|█████████▌ | 3977/10000 [07:01<07:20, 13.68it/s] Update steps: 40%|█████████▌ | 3979/10000 [07:01<07:21, 13.63it/s] Update steps: 40%|█████████▌ | 3981/10000 [07:01<07:19, 13.68it/s] Update steps: 40%|█████████▌ | 3983/10000 [07:02<07:24, 13.53it/s] Update steps: 40%|█████████▌ | 3985/10000 [07:02<07:24, 13.52it/s] Update steps: 40%|█████████▌ | 3987/10000 [07:02<07:18, 13.70it/s] Update steps: 40%|█████████▌ | 3989/10000 [07:02<07:18, 13.71it/s] Update steps: 40%|█████████▌ | 3991/10000 [07:02<07:17, 13.72it/s] Update steps: 40%|█████████▌ | 3993/10000 [07:02<07:15, 13.78it/s] Update steps: 40%|█████████▌ | 3995/10000 [07:02<07:19, 13.66it/s] Update steps: 40%|█████████▌ | 3997/10000 [07:03<07:19, 13.67it/s] Update steps: 40%|█████████▌ | 3999/10000 [07:03<07:18, 13.69it/s]2026-04-17 11:00:23.725 | INFO | __main__:main:490 - Performing evaluation at step 4000 +2026-04-17 11:00:23.730 | INFO | utils.eval:evaluate_model:21 - Loading validation data from 8 local validation files +2026-04-17 11:00:24.465 | INFO | utils.eval:evaluate_model:43 - Loaded validation dataset in 0.74 seconds +2026-04-17 11:00:24.465 | INFO | utils.eval:evaluate_model:60 - Eval set prepared in 0.74 seconds + Update steps: 40%|█████████▌ | 4000/10000 [07:16<07:18, 13.69it/s]2026-04-17 11:01:00.023 | INFO | __main__:main:505 - Eval loss at step 4000: 3.6203424027166866, Eval perplexity: 37.350354495681366 + Update steps: 40%|████████▊ | 4001/10000 [07:39<9:11:36, 5.52s/it] Update steps: 40%|████████▊ | 4003/10000 [07:39<6:28:07, 3.88s/it] Update steps: 40%|████████▊ | 4005/10000 [07:39<4:33:45, 2.74s/it] Update steps: 40%|████████▊ | 4007/10000 [07:40<3:13:44, 1.94s/it] Update steps: 40%|████████▊ | 4009/10000 [07:40<2:17:44, 1.38s/it] Update steps: 40%|████████▊ | 4011/10000 [07:40<1:38:34, 1.01it/s] Update steps: 40%|████████▊ | 4013/10000 [07:40<1:11:06, 1.40it/s] Update steps: 40%|█████████▋ | 4015/10000 [07:40<51:52, 1.92it/s] Update steps: 40%|█████████▋ | 4017/10000 [07:40<38:28, 2.59it/s] Update steps: 40%|█████████▋ | 4019/10000 [07:40<29:04, 3.43it/s] Update steps: 40%|█████████▋ | 4021/10000 [07:41<22:29, 4.43it/s] Update steps: 40%|█████████▋ | 4023/10000 [07:41<17:53, 5.57it/s] Update steps: 40%|█████████▋ | 4025/10000 [07:41<14:39, 6.80it/s] Update steps: 40%|█████████▋ | 4027/10000 [07:41<12:25, 8.01it/s] Update steps: 40%|█████████▋ | 4029/10000 [07:41<10:52, 9.15it/s] Update steps: 40%|█████████▋ | 4031/10000 [07:41<09:49, 10.12it/s] Update steps: 40%|█████████▋ | 4033/10000 [07:42<08:59, 11.05it/s] Update steps: 40%|█████████▋ | 4035/10000 [07:42<08:28, 11.74it/s] Update steps: 40%|█████████▋ | 4037/10000 [07:42<08:03, 12.34it/s] Update steps: 40%|█████████▋ | 4039/10000 [07:42<07:46, 12.79it/s] Update steps: 40%|█████████▋ | 4041/10000 [07:42<07:33, 13.14it/s] Update steps: 40%|█████████▋ | 4043/10000 [07:42<07:28, 13.28it/s] Update steps: 40%|█████████▋ | 4045/10000 [07:42<07:25, 13.38it/s] Update steps: 40%|█████████▋ | 4047/10000 [07:43<07:16, 13.63it/s] Update steps: 40%|█████████▋ | 4049/10000 [07:43<07:12, 13.76it/s] Update steps: 41%|█████████▋ | 4051/10000 [07:43<07:15, 13.67it/s] Update steps: 41%|█████████▋ | 4053/10000 [07:43<07:14, 13.67it/s] Update steps: 41%|█████████▋ | 4055/10000 [07:43<07:18, 13.56it/s] Update steps: 41%|█████████▋ | 4057/10000 [07:43<07:13, 13.71it/s] Update steps: 41%|█████████▋ | 4059/10000 [07:43<07:15, 13.63it/s] Update steps: 41%|█████████▋ | 4061/10000 [07:44<07:11, 13.77it/s] Update steps: 41%|█████████▊ | 4063/10000 [07:44<07:08, 13.84it/s] Update steps: 41%|█████████▊ | 4065/10000 [07:44<07:10, 13.77it/s] Update steps: 41%|█████████▊ | 4067/10000 [07:44<07:13, 13.70it/s] Update steps: 41%|█████████▊ | 4069/10000 [07:44<07:16, 13.60it/s] Update steps: 41%|█████████▊ | 4071/10000 [07:44<07:16, 13.59it/s] Update steps: 41%|█████████▊ | 4073/10000 [07:44<07:12, 13.70it/s] Update steps: 41%|█████████▊ | 4075/10000 [07:45<07:12, 13.70it/s] Update steps: 41%|█████████▊ | 4077/10000 [07:45<07:11, 13.74it/s] Update steps: 41%|█████████▊ | 4079/10000 [07:45<07:07, 13.85it/s] Update steps: 41%|█████████▊ | 4081/10000 [07:45<07:10, 13.76it/s] Update steps: 41%|█████████▊ | 4083/10000 [07:45<07:12, 13.70it/s] Update steps: 41%|█████████▊ | 4085/10000 [07:45<07:10, 13.74it/s] Update steps: 41%|█████████▊ | 4087/10000 [07:45<07:08, 13.79it/s] Update steps: 41%|█████████▊ | 4089/10000 [07:46<07:07, 13.82it/s] Update steps: 41%|█████████▊ | 4091/10000 [07:46<07:07, 13.81it/s] Update steps: 41%|█████████▊ | 4093/10000 [07:46<07:05, 13.89it/s] Update steps: 41%|█████████▊ | 4095/10000 [07:46<07:06, 13.85it/s] Update steps: 41%|█████████▊ | 4097/10000 [07:46<07:08, 13.78it/s] Update steps: 41%|█████████▊ | 4099/10000 [07:46<07:10, 13.69it/s] Update steps: 41%|█████████▊ | 4101/10000 [07:46<07:11, 13.68it/s] Update steps: 41%|█████████▊ | 4103/10000 [07:47<07:07, 13.78it/s] Update steps: 41%|█████████▊ | 4105/10000 [07:47<07:09, 13.71it/s] Update steps: 41%|█████████▊ | 4107/10000 [07:47<07:12, 13.62it/s] Update steps: 41%|█████████▊ | 4109/10000 [07:47<07:11, 13.67it/s] Update steps: 41%|█████████▊ | 4111/10000 [07:47<07:11, 13.65it/s] Update steps: 41%|█████████▊ | 4113/10000 [07:47<07:08, 13.75it/s] Update steps: 41%|█████████▉ | 4115/10000 [07:47<07:07, 13.76it/s] Update steps: 41%|█████████▉ | 4117/10000 [07:48<07:09, 13.69it/s] Update steps: 41%|█████████▉ | 4119/10000 [07:48<07:12, 13.60it/s] Update steps: 41%|█████████▉ | 4121/10000 [07:48<07:12, 13.61it/s] Update steps: 41%|█████████▉ | 4123/10000 [07:48<07:08, 13.70it/s] Update steps: 41%|█████████▉ | 4125/10000 [07:48<07:15, 13.48it/s] Update steps: 41%|█████████▉ | 4127/10000 [07:48<07:16, 13.44it/s] Update steps: 41%|█████████▉ | 4129/10000 [07:49<07:19, 13.35it/s] Update steps: 41%|█████████▉ | 4131/10000 [07:49<07:18, 13.38it/s] Update steps: 41%|█████████▉ | 4133/10000 [07:49<07:20, 13.33it/s] Update steps: 41%|█████████▉ | 4135/10000 [07:49<07:34, 12.91it/s] Update steps: 41%|█████████▉ | 4137/10000 [07:49<07:26, 13.14it/s] Update steps: 41%|█████████▉ | 4139/10000 [07:49<07:22, 13.24it/s] Update steps: 41%|█████████▉ | 4141/10000 [07:49<07:19, 13.32it/s] Update steps: 41%|█████████▉ | 4143/10000 [07:50<07:17, 13.38it/s] Update steps: 41%|█████████▉ | 4145/10000 [07:50<07:19, 13.33it/s] Update steps: 41%|█████████▉ | 4147/10000 [07:50<07:16, 13.41it/s] Update steps: 41%|█████████▉ | 4149/10000 [07:50<07:13, 13.49it/s] Update steps: 42%|█████████▉ | 4151/10000 [07:50<07:10, 13.60it/s] Update steps: 42%|█████████▉ | 4153/10000 [07:50<07:13, 13.49it/s] Update steps: 42%|█████████▉ | 4155/10000 [07:50<07:07, 13.67it/s] Update steps: 42%|█████████▉ | 4157/10000 [07:51<07:12, 13.52it/s] Update steps: 42%|█████████▉ | 4159/10000 [07:51<07:08, 13.64it/s] Update steps: 42%|█████████▉ | 4161/10000 [07:51<07:05, 13.71it/s] Update steps: 42%|█████████▉ | 4163/10000 [07:51<07:05, 13.73it/s] Update steps: 42%|█████████▉ | 4165/10000 [07:51<07:08, 13.61it/s] Update steps: 42%|██████████ | 4167/10000 [07:51<07:07, 13.64it/s] Update steps: 42%|██████████ | 4169/10000 [07:51<07:07, 13.65it/s] Update steps: 42%|██████████ | 4171/10000 [07:52<07:07, 13.62it/s] Update steps: 42%|██████████ | 4173/10000 [07:52<07:10, 13.52it/s] Update steps: 42%|██████████ | 4175/10000 [07:52<07:10, 13.53it/s] Update steps: 42%|██████████ | 4177/10000 [07:52<07:06, 13.66it/s] Update steps: 42%|██████████ | 4179/10000 [07:52<07:06, 13.66it/s] Update steps: 42%|██████████ | 4181/10000 [07:52<07:09, 13.55it/s] Update steps: 42%|██████████ | 4183/10000 [07:53<07:09, 13.55it/s] Update steps: 42%|██████████ | 4185/10000 [07:53<07:09, 13.53it/s] Update steps: 42%|██████████ | 4187/10000 [07:53<07:07, 13.59it/s] Update steps: 42%|██████████ | 4189/10000 [07:53<07:03, 13.72it/s] Update steps: 42%|██████████ | 4191/10000 [07:53<07:00, 13.82it/s] Update steps: 42%|██████████ | 4193/10000 [07:53<07:09, 13.51it/s] Update steps: 42%|██████████ | 4195/10000 [07:53<07:04, 13.67it/s] Update steps: 42%|██████████ | 4197/10000 [07:54<07:01, 13.76it/s] Update steps: 42%|██████████ | 4199/10000 [07:54<07:03, 13.70it/s] Update steps: 42%|██████████ | 4201/10000 [07:54<07:03, 13.71it/s] Update steps: 42%|██████████ | 4203/10000 [07:54<07:02, 13.73it/s] Update steps: 42%|██████████ | 4205/10000 [07:54<06:59, 13.80it/s] Update steps: 42%|██████████ | 4207/10000 [07:54<07:01, 13.74it/s] Update steps: 42%|██████████ | 4209/10000 [07:54<06:58, 13.83it/s] Update steps: 42%|██████████ | 4211/10000 [07:55<06:58, 13.84it/s] Update steps: 42%|██████████ | 4213/10000 [07:55<07:00, 13.76it/s] Update steps: 42%|██████████ | 4215/10000 [07:55<07:05, 13.61it/s] Update steps: 42%|██████████ | 4217/10000 [07:55<07:04, 13.62it/s] Update steps: 42%|██████████▏ | 4219/10000 [07:55<07:04, 13.61it/s] Update steps: 42%|██████████▏ | 4221/10000 [07:55<07:05, 13.57it/s] Update steps: 42%|██████████▏ | 4223/10000 [07:55<07:01, 13.70it/s] Update steps: 42%|██████████▏ | 4225/10000 [07:56<07:00, 13.74it/s] Update steps: 42%|██████████▏ | 4227/10000 [07:56<06:58, 13.81it/s] Update steps: 42%|██████████▏ | 4229/10000 [07:56<06:59, 13.77it/s] Update steps: 42%|██████████▏ | 4231/10000 [07:56<06:59, 13.76it/s] Update steps: 42%|██████████▏ | 4233/10000 [07:56<06:58, 13.78it/s] Update steps: 42%|██████████▏ | 4235/10000 [07:56<06:57, 13.81it/s] Update steps: 42%|██████████▏ | 4237/10000 [07:56<07:03, 13.62it/s] Update steps: 42%|██████████▏ | 4239/10000 [07:57<07:02, 13.64it/s] Update steps: 42%|██████████▏ | 4241/10000 [07:57<07:01, 13.65it/s] Update steps: 42%|██████████▏ | 4243/10000 [07:57<07:03, 13.59it/s] Update steps: 42%|██████████▏ | 4245/10000 [07:57<07:03, 13.59it/s] Update steps: 42%|██████████▏ | 4247/10000 [07:57<07:03, 13.58it/s] Update steps: 42%|██████████▏ | 4249/10000 [07:57<06:59, 13.70it/s] Update steps: 43%|██████████▏ | 4251/10000 [07:57<06:57, 13.79it/s] Update steps: 43%|██████████▏ | 4253/10000 [07:58<07:00, 13.68it/s] Update steps: 43%|██████████▏ | 4255/10000 [07:58<06:58, 13.72it/s] Update steps: 43%|██████████▏ | 4257/10000 [07:58<06:59, 13.69it/s] Update steps: 43%|██████████▏ | 4259/10000 [07:58<06:57, 13.74it/s] Update steps: 43%|██████████▏ | 4261/10000 [07:58<07:02, 13.59it/s] Update steps: 43%|██████████▏ | 4263/10000 [07:58<07:01, 13.62it/s] Update steps: 43%|██████████▏ | 4265/10000 [07:58<07:01, 13.60it/s] Update steps: 43%|██████████▏ | 4267/10000 [07:59<07:00, 13.62it/s] Update steps: 43%|██████████▏ | 4269/10000 [07:59<07:00, 13.61it/s] Update steps: 43%|██████████▎ | 4271/10000 [07:59<07:03, 13.52it/s] Update steps: 43%|██████████▎ | 4273/10000 [07:59<07:01, 13.58it/s] Update steps: 43%|██████████▎ | 4275/10000 [07:59<07:02, 13.56it/s] Update steps: 43%|██████████▎ | 4277/10000 [07:59<07:00, 13.59it/s] Update steps: 43%|██████████▎ | 4279/10000 [08:00<07:01, 13.57it/s] Update steps: 43%|██████████▎ | 4281/10000 [08:00<06:58, 13.67it/s] Update steps: 43%|██████████▎ | 4283/10000 [08:00<06:58, 13.65it/s] Update steps: 43%|██████████▎ | 4285/10000 [08:00<06:59, 13.61it/s] Update steps: 43%|██████████▎ | 4287/10000 [08:00<07:01, 13.54it/s] Update steps: 43%|██████████▎ | 4289/10000 [08:00<07:01, 13.56it/s] Update steps: 43%|██████████▎ | 4291/10000 [08:00<07:02, 13.51it/s] Update steps: 43%|██████████▎ | 4293/10000 [08:01<06:59, 13.60it/s] Update steps: 43%|██████████▎ | 4295/10000 [08:01<07:01, 13.53it/s] Update steps: 43%|██████████▎ | 4297/10000 [08:01<07:02, 13.49it/s] Update steps: 43%|██████████▎ | 4299/10000 [08:01<07:03, 13.47it/s] Update steps: 43%|██████████▎ | 4301/10000 [08:01<07:02, 13.50it/s] Update steps: 43%|██████████▎ | 4303/10000 [08:01<06:58, 13.60it/s] Update steps: 43%|██████████▎ | 4305/10000 [08:01<07:00, 13.55it/s] Update steps: 43%|██████████▎ | 4307/10000 [08:02<07:00, 13.54it/s] Update steps: 43%|██████████▎ | 4309/10000 [08:02<06:58, 13.59it/s] Update steps: 43%|██████████▎ | 4311/10000 [08:02<06:53, 13.74it/s] Update steps: 43%|██████████▎ | 4313/10000 [08:02<07:00, 13.54it/s] Update steps: 43%|██████████▎ | 4315/10000 [08:02<06:58, 13.60it/s] Update steps: 43%|██████████▎ | 4317/10000 [08:02<07:04, 13.40it/s] Update steps: 43%|██████████▎ | 4319/10000 [08:02<07:03, 13.43it/s] Update steps: 43%|██████████▎ | 4321/10000 [08:03<06:59, 13.53it/s] Update steps: 43%|██████████▍ | 4323/10000 [08:03<06:57, 13.58it/s] Update steps: 43%|██████████▍ | 4325/10000 [08:03<06:56, 13.62it/s] Update steps: 43%|██████████▍ | 4327/10000 [08:03<06:55, 13.65it/s] Update steps: 43%|██████████▍ | 4329/10000 [08:03<06:53, 13.72it/s] Update steps: 43%|██████████▍ | 4331/10000 [08:03<06:50, 13.82it/s] Update steps: 43%|██████████▍ | 4333/10000 [08:04<06:53, 13.71it/s] Update steps: 43%|██████████▍ | 4335/10000 [08:04<07:03, 13.37it/s] Update steps: 43%|██████████▍ | 4337/10000 [08:04<06:59, 13.50it/s] Update steps: 43%|██████████▍ | 4339/10000 [08:04<06:55, 13.61it/s] Update steps: 43%|██████████▍ | 4341/10000 [08:04<06:55, 13.63it/s] Update steps: 43%|██████████▍ | 4343/10000 [08:04<06:55, 13.63it/s] Update steps: 43%|██████████▍ | 4345/10000 [08:04<06:53, 13.67it/s] Update steps: 43%|██████████▍ | 4347/10000 [08:05<06:52, 13.72it/s] Update steps: 43%|██████████▍ | 4349/10000 [08:05<06:50, 13.75it/s] Update steps: 44%|██████████▍ | 4351/10000 [08:05<06:50, 13.74it/s] Update steps: 44%|██████████▍ | 4353/10000 [08:05<06:49, 13.79it/s] Update steps: 44%|██████████▍ | 4355/10000 [08:05<06:50, 13.77it/s] Update steps: 44%|██████████▍ | 4357/10000 [08:05<06:48, 13.80it/s] Update steps: 44%|██████████▍ | 4359/10000 [08:05<06:48, 13.80it/s] Update steps: 44%|██████████▍ | 4361/10000 [08:06<06:48, 13.80it/s] Update steps: 44%|██████████▍ | 4363/10000 [08:06<06:46, 13.86it/s] Update steps: 44%|██████████▍ | 4365/10000 [08:06<06:46, 13.88it/s] Update steps: 44%|██████████▍ | 4367/10000 [08:06<06:46, 13.87it/s] Update steps: 44%|██████████▍ | 4369/10000 [08:06<06:47, 13.81it/s] Update steps: 44%|██████████▍ | 4371/10000 [08:06<06:49, 13.74it/s] Update steps: 44%|██████████▍ | 4373/10000 [08:06<06:52, 13.66it/s] Update steps: 44%|██████████▌ | 4375/10000 [08:07<06:52, 13.65it/s] Update steps: 44%|██████████▌ | 4377/10000 [08:07<06:53, 13.60it/s] Update steps: 44%|██████████▌ | 4379/10000 [08:07<06:51, 13.66it/s] Update steps: 44%|██████████▌ | 4381/10000 [08:07<06:51, 13.67it/s] Update steps: 44%|██████████▌ | 4383/10000 [08:07<06:49, 13.72it/s] Update steps: 44%|██████████▌ | 4385/10000 [08:07<06:46, 13.81it/s] Update steps: 44%|██████████▌ | 4387/10000 [08:07<06:47, 13.76it/s] Update steps: 44%|██████████▌ | 4389/10000 [08:08<06:46, 13.81it/s] Update steps: 44%|██████████▌ | 4391/10000 [08:08<06:51, 13.64it/s] Update steps: 44%|██████████▌ | 4393/10000 [08:08<06:53, 13.54it/s] Update steps: 44%|██████████▌ | 4395/10000 [08:08<06:55, 13.50it/s] Update steps: 44%|██████████▌ | 4397/10000 [08:08<06:49, 13.68it/s] Update steps: 44%|██████████▌ | 4399/10000 [08:08<06:52, 13.58it/s] Update steps: 44%|██████████▌ | 4401/10000 [08:08<06:49, 13.66it/s] Update steps: 44%|██████████▌ | 4403/10000 [08:09<06:47, 13.74it/s] Update steps: 44%|██████████▌ | 4405/10000 [08:09<06:49, 13.66it/s] Update steps: 44%|██████████▌ | 4407/10000 [08:09<06:48, 13.70it/s] Update steps: 44%|██████████▌ | 4409/10000 [08:09<06:46, 13.74it/s] Update steps: 44%|██████████▌ | 4411/10000 [08:09<06:49, 13.65it/s] Update steps: 44%|██████████▌ | 4413/10000 [08:09<06:51, 13.59it/s] Update steps: 44%|██████████▌ | 4415/10000 [08:09<06:49, 13.62it/s] Update steps: 44%|██████████▌ | 4417/10000 [08:10<07:00, 13.29it/s] Update steps: 44%|██████████▌ | 4419/10000 [08:10<07:00, 13.27it/s] Update steps: 44%|██████████▌ | 4421/10000 [08:10<06:53, 13.48it/s] Update steps: 44%|██████████▌ | 4423/10000 [08:10<06:51, 13.54it/s] Update steps: 44%|██████████▌ | 4425/10000 [08:10<06:53, 13.49it/s] Update steps: 44%|██████████▌ | 4427/10000 [08:10<06:53, 13.47it/s] Update steps: 44%|██████████▋ | 4429/10000 [08:11<06:51, 13.52it/s] Update steps: 44%|██████████▋ | 4431/10000 [08:11<06:50, 13.57it/s] Update steps: 44%|██████████▋ | 4433/10000 [08:11<06:52, 13.50it/s] Update steps: 44%|██████████▋ | 4435/10000 [08:11<06:53, 13.45it/s] Update steps: 44%|██████████▋ | 4437/10000 [08:11<06:51, 13.52it/s] Update steps: 44%|██████████▋ | 4439/10000 [08:11<06:49, 13.57it/s] Update steps: 44%|██████████▋ | 4441/10000 [08:11<06:47, 13.64it/s] Update steps: 44%|██████████▋ | 4443/10000 [08:12<06:49, 13.58it/s] Update steps: 44%|██████████▋ | 4445/10000 [08:12<06:45, 13.70it/s] Update steps: 44%|██████████▋ | 4447/10000 [08:12<06:44, 13.74it/s] Update steps: 44%|██████████▋ | 4449/10000 [08:12<06:45, 13.69it/s] Update steps: 45%|██████████▋ | 4451/10000 [08:12<06:45, 13.67it/s] Update steps: 45%|██████████▋ | 4453/10000 [08:12<06:43, 13.75it/s] Update steps: 45%|██████████▋ | 4455/10000 [08:12<06:47, 13.61it/s] Update steps: 45%|██████████▋ | 4457/10000 [08:13<06:45, 13.66it/s] Update steps: 45%|██████████▋ | 4459/10000 [08:13<06:45, 13.67it/s] Update steps: 45%|██████████▋ | 4461/10000 [08:13<06:43, 13.73it/s] Update steps: 45%|██████████▋ | 4463/10000 [08:13<06:41, 13.80it/s] Update steps: 45%|██████████▋ | 4465/10000 [08:13<06:42, 13.74it/s] Update steps: 45%|██████████▋ | 4467/10000 [08:13<06:42, 13.76it/s] Update steps: 45%|██████████▋ | 4469/10000 [08:13<06:40, 13.81it/s] Update steps: 45%|██████████▋ | 4471/10000 [08:14<06:44, 13.67it/s] Update steps: 45%|██████████▋ | 4473/10000 [08:14<06:43, 13.71it/s] Update steps: 45%|██████████▋ | 4475/10000 [08:14<06:44, 13.66it/s] Update steps: 45%|██████████▋ | 4477/10000 [08:14<06:43, 13.68it/s] Update steps: 45%|██████████▋ | 4479/10000 [08:14<06:46, 13.57it/s] Update steps: 45%|██████████▊ | 4481/10000 [08:14<06:45, 13.60it/s] Update steps: 45%|██████████▊ | 4483/10000 [08:14<06:48, 13.52it/s] Update steps: 45%|██████████▊ | 4485/10000 [08:15<06:43, 13.67it/s] Update steps: 45%|██████████▊ | 4487/10000 [08:15<06:41, 13.73it/s] Update steps: 45%|██████████▊ | 4489/10000 [08:15<06:42, 13.69it/s] Update steps: 45%|██████████▊ | 4491/10000 [08:15<06:41, 13.71it/s] Update steps: 45%|██████████▊ | 4493/10000 [08:15<06:41, 13.71it/s] Update steps: 45%|██████████▊ | 4495/10000 [08:15<06:40, 13.74it/s] Update steps: 45%|██████████▊ | 4497/10000 [08:16<06:39, 13.79it/s] Update steps: 45%|██████████▊ | 4499/10000 [08:16<06:40, 13.74it/s] Update steps: 45%|██████████▊ | 4501/10000 [08:16<06:37, 13.82it/s] Update steps: 45%|██████████▊ | 4503/10000 [08:16<06:37, 13.84it/s] Update steps: 45%|██████████▊ | 4505/10000 [08:16<06:35, 13.91it/s] Update steps: 45%|██████████▊ | 4507/10000 [08:16<06:34, 13.93it/s] Update steps: 45%|██████████▊ | 4509/10000 [08:16<06:35, 13.90it/s] Update steps: 45%|██████████▊ | 4511/10000 [08:17<06:39, 13.75it/s] Update steps: 45%|██████████▊ | 4513/10000 [08:17<06:38, 13.77it/s] Update steps: 45%|██████████▊ | 4515/10000 [08:17<06:40, 13.69it/s] Update steps: 45%|██████████▊ | 4517/10000 [08:17<06:41, 13.67it/s] Update steps: 45%|██████████▊ | 4519/10000 [08:17<06:41, 13.64it/s] Update steps: 45%|██████████▊ | 4521/10000 [08:17<06:44, 13.54it/s] Update steps: 45%|██████████▊ | 4523/10000 [08:17<06:43, 13.57it/s] Update steps: 45%|██████████▊ | 4525/10000 [08:18<06:45, 13.52it/s] Update steps: 45%|██████████▊ | 4527/10000 [08:18<06:46, 13.48it/s] Update steps: 45%|██████████▊ | 4529/10000 [08:18<06:47, 13.42it/s] Update steps: 45%|██████████▊ | 4531/10000 [08:18<06:45, 13.47it/s] Update steps: 45%|██████████▉ | 4533/10000 [08:18<06:42, 13.59it/s] Update steps: 45%|██████████▉ | 4535/10000 [08:18<06:42, 13.59it/s] Update steps: 45%|██████████▉ | 4537/10000 [08:18<06:47, 13.40it/s] Update steps: 45%|██████████▉ | 4539/10000 [08:19<06:44, 13.49it/s] Update steps: 45%|██████████▉ | 4541/10000 [08:19<06:42, 13.55it/s] Update steps: 45%|██████████▉ | 4543/10000 [08:19<06:38, 13.68it/s] Update steps: 45%|██████████▉ | 4545/10000 [08:19<06:36, 13.75it/s] Update steps: 45%|██████████▉ | 4547/10000 [08:19<06:34, 13.81it/s] Update steps: 45%|██████████▉ | 4549/10000 [08:19<06:38, 13.69it/s] Update steps: 46%|██████████▉ | 4551/10000 [08:19<06:35, 13.77it/s] Update steps: 46%|██████████▉ | 4553/10000 [08:20<06:34, 13.81it/s] Update steps: 46%|██████████▉ | 4555/10000 [08:20<06:37, 13.68it/s] Update steps: 46%|██████████▉ | 4557/10000 [08:20<06:40, 13.60it/s] Update steps: 46%|██████████▉ | 4559/10000 [08:20<06:37, 13.68it/s] Update steps: 46%|██████████▉ | 4561/10000 [08:20<06:35, 13.74it/s] Update steps: 46%|██████████▉ | 4563/10000 [08:20<06:33, 13.80it/s] Update steps: 46%|██████████▉ | 4565/10000 [08:20<06:32, 13.84it/s] Update steps: 46%|██████████▉ | 4567/10000 [08:21<06:31, 13.87it/s] Update steps: 46%|██████████▉ | 4569/10000 [08:21<06:35, 13.73it/s] Update steps: 46%|██████████▉ | 4571/10000 [08:21<06:37, 13.65it/s] Update steps: 46%|██████████▉ | 4573/10000 [08:21<06:41, 13.53it/s] Update steps: 46%|██████████▉ | 4575/10000 [08:21<06:41, 13.51it/s] Update steps: 46%|██████████▉ | 4577/10000 [08:21<06:37, 13.63it/s] Update steps: 46%|██████████▉ | 4579/10000 [08:22<06:35, 13.70it/s] Update steps: 46%|██████████▉ | 4581/10000 [08:22<06:34, 13.74it/s] Update steps: 46%|██████████▉ | 4583/10000 [08:22<06:32, 13.81it/s] Update steps: 46%|███████████ | 4585/10000 [08:22<06:31, 13.82it/s] Update steps: 46%|███████████ | 4587/10000 [08:22<06:29, 13.89it/s] Update steps: 46%|███████████ | 4589/10000 [08:22<06:33, 13.76it/s] Update steps: 46%|███████████ | 4591/10000 [08:22<06:29, 13.88it/s] Update steps: 46%|███████████ | 4593/10000 [08:23<06:32, 13.79it/s] Update steps: 46%|███████████ | 4595/10000 [08:23<06:42, 13.42it/s] Update steps: 46%|███████████ | 4597/10000 [08:23<06:38, 13.54it/s] Update steps: 46%|███████████ | 4599/10000 [08:23<06:37, 13.60it/s] Update steps: 46%|███████████ | 4601/10000 [08:23<06:36, 13.61it/s] Update steps: 46%|███████████ | 4603/10000 [08:23<06:33, 13.71it/s] Update steps: 46%|███████████ | 4605/10000 [08:23<06:29, 13.84it/s] Update steps: 46%|███████████ | 4607/10000 [08:24<06:30, 13.80it/s] Update steps: 46%|███████████ | 4609/10000 [08:24<06:33, 13.72it/s] Update steps: 46%|███████████ | 4611/10000 [08:24<06:33, 13.69it/s] Update steps: 46%|███████████ | 4613/10000 [08:24<06:30, 13.79it/s] Update steps: 46%|███████████ | 4615/10000 [08:24<06:32, 13.71it/s] Update steps: 46%|███████████ | 4617/10000 [08:24<06:29, 13.82it/s] Update steps: 46%|███████████ | 4619/10000 [08:24<06:30, 13.77it/s] Update steps: 46%|███████████ | 4621/10000 [08:25<06:29, 13.81it/s] Update steps: 46%|███████████ | 4623/10000 [08:25<06:27, 13.88it/s] Update steps: 46%|███████████ | 4625/10000 [08:25<06:32, 13.70it/s] Update steps: 46%|███████████ | 4627/10000 [08:25<06:32, 13.69it/s] Update steps: 46%|███████████ | 4629/10000 [08:25<06:28, 13.81it/s] Update steps: 46%|███████████ | 4631/10000 [08:25<06:28, 13.80it/s] Update steps: 46%|███████████ | 4633/10000 [08:25<06:26, 13.88it/s] Update steps: 46%|███████████ | 4635/10000 [08:26<06:27, 13.85it/s] Update steps: 46%|███████████▏ | 4637/10000 [08:26<06:27, 13.83it/s] Update steps: 46%|███████████▏ | 4639/10000 [08:26<06:26, 13.88it/s] Update steps: 46%|███████████▏ | 4641/10000 [08:26<06:28, 13.80it/s] Update steps: 46%|███████████▏ | 4643/10000 [08:26<06:30, 13.70it/s] Update steps: 46%|███████████▏ | 4645/10000 [08:26<06:28, 13.77it/s] Update steps: 46%|███████████▏ | 4647/10000 [08:26<06:27, 13.83it/s] Update steps: 46%|███████████▏ | 4649/10000 [08:27<06:29, 13.74it/s] Update steps: 47%|███████████▏ | 4651/10000 [08:27<06:29, 13.73it/s] Update steps: 47%|███████████▏ | 4653/10000 [08:27<06:31, 13.67it/s] Update steps: 47%|███████████▏ | 4655/10000 [08:27<06:33, 13.59it/s] Update steps: 47%|███████████▏ | 4657/10000 [08:27<06:31, 13.64it/s] Update steps: 47%|███████████▏ | 4659/10000 [08:27<06:30, 13.69it/s] Update steps: 47%|███████████▏ | 4661/10000 [08:27<06:29, 13.71it/s] Update steps: 47%|███████████▏ | 4663/10000 [08:28<06:29, 13.71it/s] Update steps: 47%|███████████▏ | 4665/10000 [08:28<06:30, 13.67it/s] Update steps: 47%|███████████▏ | 4667/10000 [08:28<06:29, 13.70it/s] Update steps: 47%|███████████▏ | 4669/10000 [08:28<06:26, 13.80it/s] Update steps: 47%|███████████▏ | 4671/10000 [08:28<06:25, 13.82it/s] Update steps: 47%|███████████▏ | 4673/10000 [08:28<06:26, 13.79it/s] Update steps: 47%|███████████▏ | 4675/10000 [08:28<06:25, 13.82it/s] Update steps: 47%|███████████▏ | 4677/10000 [08:29<06:24, 13.83it/s] Update steps: 47%|███████████▏ | 4679/10000 [08:29<06:28, 13.71it/s] Update steps: 47%|███████████▏ | 4681/10000 [08:29<06:26, 13.75it/s] Update steps: 47%|███████████▏ | 4683/10000 [08:29<06:26, 13.75it/s] Update steps: 47%|███████████▏ | 4685/10000 [08:29<06:24, 13.83it/s] Update steps: 47%|███████████▏ | 4687/10000 [08:29<06:24, 13.82it/s] Update steps: 47%|███████████▎ | 4689/10000 [08:30<06:25, 13.79it/s] Update steps: 47%|███████████▎ | 4691/10000 [08:30<06:27, 13.69it/s] Update steps: 47%|███████████▎ | 4693/10000 [08:30<06:25, 13.76it/s] Update steps: 47%|███████████▎ | 4695/10000 [08:30<06:25, 13.77it/s] Update steps: 47%|███████████▎ | 4697/10000 [08:30<06:23, 13.81it/s] Update steps: 47%|███████████▎ | 4699/10000 [08:30<06:25, 13.77it/s] Update steps: 47%|███████████▎ | 4701/10000 [08:30<06:25, 13.75it/s] Update steps: 47%|███████████▎ | 4703/10000 [08:31<06:27, 13.68it/s] Update steps: 47%|███████████▎ | 4705/10000 [08:31<06:25, 13.72it/s] Update steps: 47%|███████████▎ | 4707/10000 [08:31<06:25, 13.72it/s] Update steps: 47%|███████████▎ | 4709/10000 [08:31<06:27, 13.67it/s] Update steps: 47%|███████████▎ | 4711/10000 [08:31<06:26, 13.69it/s] Update steps: 47%|███████████▎ | 4713/10000 [08:31<06:23, 13.79it/s] Update steps: 47%|███████████▎ | 4715/10000 [08:31<06:34, 13.39it/s] Update steps: 47%|███████████▎ | 4717/10000 [08:32<06:31, 13.49it/s] Update steps: 47%|███████████▎ | 4719/10000 [08:32<06:29, 13.56it/s] Update steps: 47%|███████████▎ | 4721/10000 [08:32<06:28, 13.57it/s] Update steps: 47%|███████████▎ | 4723/10000 [08:32<06:29, 13.56it/s] Update steps: 47%|███████████▎ | 4725/10000 [08:32<06:26, 13.65it/s] Update steps: 47%|███████████▎ | 4727/10000 [08:32<06:23, 13.77it/s] Update steps: 47%|███████████▎ | 4729/10000 [08:32<06:20, 13.85it/s] Update steps: 47%|███████████▎ | 4731/10000 [08:33<06:20, 13.83it/s] Update steps: 47%|███████████▎ | 4733/10000 [08:33<06:22, 13.79it/s] Update steps: 47%|███████████▎ | 4735/10000 [08:33<06:26, 13.63it/s] Update steps: 47%|███████████▎ | 4737/10000 [08:33<06:24, 13.67it/s] Update steps: 47%|███████████▎ | 4739/10000 [08:33<06:25, 13.64it/s] Update steps: 47%|███████████▍ | 4741/10000 [08:33<06:23, 13.70it/s] Update steps: 47%|███████████▍ | 4743/10000 [08:33<06:22, 13.75it/s] Update steps: 47%|███████████▍ | 4745/10000 [08:34<06:24, 13.66it/s] Update steps: 47%|███████████▍ | 4747/10000 [08:34<06:28, 13.54it/s] Update steps: 47%|███████████▍ | 4749/10000 [08:34<06:25, 13.60it/s] Update steps: 48%|███████████▍ | 4751/10000 [08:34<06:24, 13.67it/s] Update steps: 48%|███████████▍ | 4753/10000 [08:34<06:22, 13.71it/s] Update steps: 48%|███████████▍ | 4755/10000 [08:34<06:21, 13.74it/s] Update steps: 48%|███████████▍ | 4757/10000 [08:34<06:28, 13.51it/s] Update steps: 48%|███████████▍ | 4759/10000 [08:35<06:29, 13.45it/s] Update steps: 48%|███████████▍ | 4761/10000 [08:35<06:30, 13.42it/s] Update steps: 48%|███████████▍ | 4763/10000 [08:35<06:25, 13.60it/s] Update steps: 48%|███████████▍ | 4765/10000 [08:35<06:25, 13.57it/s] Update steps: 48%|███████████▍ | 4767/10000 [08:35<06:24, 13.60it/s] Update steps: 48%|███████████▍ | 4769/10000 [08:35<06:23, 13.62it/s] Update steps: 48%|███████████▍ | 4771/10000 [08:36<06:20, 13.76it/s] Update steps: 48%|███████████▍ | 4773/10000 [08:36<06:22, 13.66it/s] Update steps: 48%|███████████▍ | 4775/10000 [08:36<06:24, 13.60it/s] Update steps: 48%|███████████▍ | 4777/10000 [08:36<06:20, 13.73it/s] Update steps: 48%|███████████▍ | 4779/10000 [08:36<06:22, 13.66it/s] Update steps: 48%|███████████▍ | 4781/10000 [08:36<06:23, 13.60it/s] Update steps: 48%|███████████▍ | 4783/10000 [08:36<06:20, 13.72it/s] Update steps: 48%|███████████▍ | 4785/10000 [08:37<06:19, 13.74it/s] Update steps: 48%|███████████▍ | 4787/10000 [08:37<06:19, 13.75it/s] Update steps: 48%|███████████▍ | 4789/10000 [08:37<06:16, 13.82it/s] Update steps: 48%|███████████▍ | 4791/10000 [08:37<06:16, 13.83it/s] Update steps: 48%|███████████▌ | 4793/10000 [08:37<06:16, 13.82it/s] Update steps: 48%|███████████▌ | 4795/10000 [08:37<06:17, 13.79it/s] Update steps: 48%|███████████▌ | 4797/10000 [08:37<06:18, 13.73it/s] Update steps: 48%|███████████▌ | 4799/10000 [08:38<06:20, 13.67it/s] Update steps: 48%|███████████▌ | 4801/10000 [08:38<06:20, 13.68it/s] Update steps: 48%|███████████▌ | 4803/10000 [08:38<06:18, 13.73it/s] Update steps: 48%|███████████▌ | 4805/10000 [08:38<06:17, 13.76it/s] Update steps: 48%|███████████▌ | 4807/10000 [08:38<06:16, 13.78it/s] Update steps: 48%|███████████▌ | 4809/10000 [08:38<06:17, 13.77it/s] Update steps: 48%|███████████▌ | 4811/10000 [08:38<06:18, 13.70it/s] Update steps: 48%|███████████▌ | 4813/10000 [08:39<06:20, 13.65it/s] Update steps: 48%|███████████▌ | 4815/10000 [08:39<06:19, 13.67it/s] Update steps: 48%|███████████▌ | 4817/10000 [08:39<06:17, 13.71it/s] Update steps: 48%|███████████▌ | 4819/10000 [08:39<06:20, 13.62it/s] Update steps: 48%|███████████▌ | 4821/10000 [08:39<06:19, 13.65it/s] Update steps: 48%|███████████▌ | 4823/10000 [08:39<06:16, 13.74it/s] Update steps: 48%|███████████▌ | 4825/10000 [08:39<06:16, 13.75it/s] Update steps: 48%|███████████▌ | 4827/10000 [08:40<06:15, 13.78it/s] Update steps: 48%|███████████▌ | 4829/10000 [08:40<06:17, 13.70it/s] Update steps: 48%|███████████▌ | 4831/10000 [08:40<06:19, 13.61it/s] Update steps: 48%|███████████▌ | 4833/10000 [08:40<06:21, 13.56it/s] Update steps: 48%|███████████▌ | 4835/10000 [08:40<06:18, 13.64it/s] Update steps: 48%|███████████▌ | 4837/10000 [08:40<06:16, 13.70it/s] Update steps: 48%|███████████▌ | 4839/10000 [08:40<06:13, 13.80it/s] Update steps: 48%|███████████▌ | 4841/10000 [08:41<06:13, 13.81it/s] Update steps: 48%|███████████▌ | 4843/10000 [08:41<06:14, 13.76it/s] Update steps: 48%|███████████▋ | 4845/10000 [08:41<06:14, 13.78it/s] Update steps: 48%|███████████▋ | 4847/10000 [08:41<06:12, 13.83it/s] Update steps: 48%|███████████▋ | 4849/10000 [08:41<06:11, 13.85it/s] Update steps: 49%|███████████▋ | 4851/10000 [08:41<06:12, 13.82it/s] Update steps: 49%|███████████▋ | 4853/10000 [08:41<06:12, 13.81it/s] Update steps: 49%|███████████▋ | 4855/10000 [08:42<06:13, 13.79it/s] Update steps: 49%|███████████▋ | 4857/10000 [08:42<06:10, 13.89it/s] Update steps: 49%|███████████▋ | 4859/10000 [08:42<06:11, 13.84it/s] Update steps: 49%|███████████▋ | 4861/10000 [08:42<06:10, 13.85it/s] Update steps: 49%|███████████▋ | 4863/10000 [08:42<06:14, 13.73it/s] Update steps: 49%|███████████▋ | 4865/10000 [08:42<06:11, 13.84it/s] Update steps: 49%|███████████▋ | 4867/10000 [08:42<06:13, 13.76it/s] Update steps: 49%|███████████▋ | 4869/10000 [08:43<06:14, 13.70it/s] Update steps: 49%|███████████▋ | 4871/10000 [08:43<06:11, 13.81it/s] Update steps: 49%|███████████▋ | 4873/10000 [08:43<06:08, 13.90it/s] Update steps: 49%|███████████▋ | 4875/10000 [08:43<06:11, 13.81it/s] Update steps: 49%|███████████▋ | 4877/10000 [08:43<06:12, 13.76it/s] Update steps: 49%|███████████▋ | 4879/10000 [08:43<06:15, 13.63it/s] Update steps: 49%|███████████▋ | 4881/10000 [08:44<06:12, 13.75it/s] Update steps: 49%|███████████▋ | 4883/10000 [08:44<06:13, 13.71it/s] Update steps: 49%|███████████▋ | 4885/10000 [08:44<06:15, 13.64it/s] Update steps: 49%|███████████▋ | 4887/10000 [08:44<06:11, 13.77it/s] Update steps: 49%|███████████▋ | 4889/10000 [08:44<06:12, 13.73it/s] Update steps: 49%|███████████▋ | 4891/10000 [08:44<06:09, 13.83it/s] Update steps: 49%|███████████▋ | 4893/10000 [08:44<06:06, 13.93it/s] Update steps: 49%|███████████▋ | 4895/10000 [08:45<06:06, 13.93it/s] Update steps: 49%|███████████▊ | 4897/10000 [08:45<06:05, 13.96it/s] Update steps: 49%|███████████▊ | 4899/10000 [08:45<06:07, 13.88it/s] Update steps: 49%|███████████▊ | 4901/10000 [08:45<06:08, 13.83it/s] Update steps: 49%|███████████▊ | 4903/10000 [08:45<06:08, 13.83it/s] Update steps: 49%|███████████▊ | 4905/10000 [08:45<06:13, 13.65it/s] Update steps: 49%|███████████▊ | 4907/10000 [08:45<06:13, 13.62it/s] Update steps: 49%|███████████▊ | 4909/10000 [08:46<06:15, 13.56it/s] Update steps: 49%|███████████▊ | 4911/10000 [08:46<06:10, 13.73it/s] Update steps: 49%|███████████▊ | 4913/10000 [08:46<06:08, 13.81it/s] Update steps: 49%|███████████▊ | 4915/10000 [08:46<06:06, 13.89it/s] Update steps: 49%|███████████▊ | 4917/10000 [08:46<06:04, 13.94it/s] Update steps: 49%|███████████▊ | 4919/10000 [08:46<06:04, 13.93it/s] Update steps: 49%|███████████▊ | 4921/10000 [08:46<06:03, 13.98it/s] Update steps: 49%|███████████▊ | 4923/10000 [08:47<06:03, 13.96it/s] Update steps: 49%|███████████▊ | 4925/10000 [08:47<06:04, 13.92it/s] Update steps: 49%|███████████▊ | 4927/10000 [08:47<06:03, 13.95it/s] Update steps: 49%|███████████▊ | 4929/10000 [08:47<06:02, 13.99it/s] Update steps: 49%|███████████▊ | 4931/10000 [08:47<06:03, 13.94it/s] Update steps: 49%|███████████▊ | 4933/10000 [08:47<06:04, 13.89it/s] Update steps: 49%|███████████▊ | 4935/10000 [08:47<06:05, 13.88it/s] Update steps: 49%|███████████▊ | 4937/10000 [08:48<06:08, 13.73it/s] Update steps: 49%|███████████▊ | 4939/10000 [08:48<06:14, 13.51it/s] Update steps: 49%|███████████▊ | 4941/10000 [08:48<06:12, 13.60it/s] Update steps: 49%|███████████▊ | 4943/10000 [08:48<06:10, 13.65it/s] Update steps: 49%|███████████▊ | 4945/10000 [08:48<06:07, 13.76it/s] Update steps: 49%|███████████▊ | 4947/10000 [08:48<06:05, 13.83it/s] Update steps: 49%|███████████▉ | 4949/10000 [08:48<06:03, 13.90it/s] Update steps: 50%|███████████▉ | 4951/10000 [08:49<06:02, 13.93it/s] Update steps: 50%|███████████▉ | 4953/10000 [08:49<06:03, 13.87it/s] Update steps: 50%|███████████▉ | 4955/10000 [08:49<06:04, 13.84it/s] Update steps: 50%|███████████▉ | 4957/10000 [08:49<06:04, 13.85it/s] Update steps: 50%|███████████▉ | 4959/10000 [08:49<06:03, 13.86it/s] Update steps: 50%|███████████▉ | 4961/10000 [08:49<06:04, 13.81it/s] Update steps: 50%|███████████▉ | 4963/10000 [08:49<06:02, 13.90it/s] Update steps: 50%|███████████▉ | 4965/10000 [08:50<06:02, 13.89it/s] Update steps: 50%|███████████▉ | 4967/10000 [08:50<06:01, 13.93it/s] Update steps: 50%|███████████▉ | 4969/10000 [08:50<05:59, 14.00it/s] Update steps: 50%|███████████▉ | 4971/10000 [08:50<05:57, 14.05it/s] Update steps: 50%|███████████▉ | 4973/10000 [08:50<06:01, 13.91it/s] Update steps: 50%|███████████▉ | 4975/10000 [08:50<06:00, 13.95it/s] Update steps: 50%|███████████▉ | 4977/10000 [08:50<06:01, 13.88it/s] Update steps: 50%|███████████▉ | 4979/10000 [08:51<06:04, 13.78it/s] Update steps: 50%|███████████▉ | 4981/10000 [08:51<06:04, 13.75it/s] Update steps: 50%|███████████▉ | 4983/10000 [08:51<06:02, 13.83it/s] Update steps: 50%|███████████▉ | 4985/10000 [08:51<06:01, 13.86it/s] Update steps: 50%|███████████▉ | 4987/10000 [08:51<06:00, 13.91it/s] Update steps: 50%|███████████▉ | 4989/10000 [08:51<06:00, 13.91it/s] Update steps: 50%|███████████▉ | 4991/10000 [08:51<05:59, 13.92it/s] Update steps: 50%|███████████▉ | 4993/10000 [08:52<05:58, 13.96it/s] Update steps: 50%|███████████▉ | 4995/10000 [08:52<05:58, 13.98it/s] Update steps: 50%|███████████▉ | 4997/10000 [08:52<05:57, 13.98it/s] Update steps: 50%|███████████▉ | 4999/10000 [08:52<06:02, 13.81it/s]2026-04-17 11:02:12.995 | INFO | __main__:main:490 - Performing evaluation at step 5000 +2026-04-17 11:02:12.999 | INFO | utils.eval:evaluate_model:21 - Loading validation data from 8 local validation files +2026-04-17 11:02:13.803 | INFO | utils.eval:evaluate_model:43 - Loaded validation dataset in 0.81 seconds +2026-04-17 11:02:13.804 | INFO | utils.eval:evaluate_model:60 - Eval set prepared in 0.81 seconds + Update steps: 50%|████████████ | 5000/10000 [09:06<06:02, 13.81it/s]2026-04-17 11:02:48.513 | INFO | __main__:main:505 - Eval loss at step 5000: 3.5570330162552777, Eval perplexity: 35.059023128491816 + Update steps: 50%|███████████ | 5001/10000 [09:28<7:29:53, 5.40s/it] Update steps: 50%|███████████ | 5003/10000 [09:28<5:16:34, 3.80s/it] Update steps: 50%|███████████ | 5005/10000 [09:28<3:43:16, 2.68s/it] Update steps: 50%|███████████ | 5007/10000 [09:28<2:38:02, 1.90s/it] Update steps: 50%|███████████ | 5009/10000 [09:28<1:52:24, 1.35s/it] Update steps: 50%|███████████ | 5011/10000 [09:28<1:20:25, 1.03it/s] Update steps: 50%|████████████ | 5013/10000 [09:29<58:07, 1.43it/s] Update steps: 50%|████████████ | 5015/10000 [09:29<42:29, 1.96it/s] Update steps: 50%|████████████ | 5017/10000 [09:29<31:29, 2.64it/s] Update steps: 50%|████████████ | 5019/10000 [09:29<23:51, 3.48it/s] Update steps: 50%|████████████ | 5021/10000 [09:29<18:30, 4.48it/s] Update steps: 50%|████████████ | 5023/10000 [09:29<14:44, 5.63it/s] Update steps: 50%|████████████ | 5025/10000 [09:29<12:09, 6.82it/s] Update steps: 50%|████████████ | 5027/10000 [09:30<10:20, 8.02it/s] Update steps: 50%|████████████ | 5029/10000 [09:30<09:01, 9.17it/s] Update steps: 50%|████████████ | 5031/10000 [09:30<08:05, 10.22it/s] Update steps: 50%|████████████ | 5033/10000 [09:30<07:27, 11.10it/s] Update steps: 50%|████████████ | 5035/10000 [09:30<06:58, 11.86it/s] Update steps: 50%|████████████ | 5037/10000 [09:30<06:37, 12.47it/s] Update steps: 50%|████████████ | 5039/10000 [09:30<06:25, 12.85it/s] Update steps: 50%|████████████ | 5041/10000 [09:31<06:18, 13.11it/s] Update steps: 50%|████████████ | 5043/10000 [09:31<06:14, 13.24it/s] Update steps: 50%|████████████ | 5045/10000 [09:31<06:08, 13.44it/s] Update steps: 50%|████████████ | 5047/10000 [09:31<06:04, 13.57it/s] Update steps: 50%|████████████ | 5049/10000 [09:31<06:04, 13.59it/s] Update steps: 51%|████████████ | 5051/10000 [09:31<06:20, 13.00it/s] Update steps: 51%|████████████▏ | 5053/10000 [09:31<06:14, 13.20it/s] Update steps: 51%|████████████▏ | 5055/10000 [09:32<06:07, 13.45it/s] Update steps: 51%|████████████▏ | 5057/10000 [09:32<06:02, 13.64it/s] Update steps: 51%|████████████▏ | 5059/10000 [09:32<06:01, 13.68it/s] Update steps: 51%|████████████▏ | 5061/10000 [09:32<05:59, 13.73it/s] Update steps: 51%|████████████▏ | 5063/10000 [09:32<05:56, 13.83it/s] Update steps: 51%|████████████▏ | 5065/10000 [09:32<05:56, 13.83it/s] Update steps: 51%|████████████▏ | 5067/10000 [09:32<05:59, 13.73it/s] Update steps: 51%|████████████▏ | 5069/10000 [09:33<05:58, 13.75it/s] Update steps: 51%|████████████▏ | 5071/10000 [09:33<05:56, 13.84it/s] Update steps: 51%|████████████▏ | 5073/10000 [09:33<05:54, 13.91it/s] Update steps: 51%|████████████▏ | 5075/10000 [09:33<05:53, 13.94it/s] Update steps: 51%|████████████▏ | 5077/10000 [09:33<05:53, 13.91it/s] Update steps: 51%|████████████▏ | 5079/10000 [09:33<05:53, 13.93it/s] Update steps: 51%|████████████▏ | 5081/10000 [09:33<05:53, 13.93it/s] Update steps: 51%|████████████▏ | 5083/10000 [09:34<05:51, 13.98it/s] Update steps: 51%|████████████▏ | 5085/10000 [09:34<05:52, 13.94it/s] Update steps: 51%|████████████▏ | 5087/10000 [09:34<05:52, 13.95it/s] Update steps: 51%|████████████▏ | 5089/10000 [09:34<05:50, 14.01it/s] Update steps: 51%|████████████▏ | 5091/10000 [09:34<05:51, 13.96it/s] Update steps: 51%|████████████▏ | 5093/10000 [09:34<05:55, 13.81it/s] Update steps: 51%|████████████▏ | 5095/10000 [09:34<05:52, 13.90it/s] Update steps: 51%|████████████▏ | 5097/10000 [09:35<05:55, 13.80it/s] Update steps: 51%|████████████▏ | 5099/10000 [09:35<05:56, 13.75it/s] Update steps: 51%|████████████▏ | 5101/10000 [09:35<05:58, 13.65it/s] Update steps: 51%|████████████▏ | 5103/10000 [09:35<05:58, 13.64it/s] Update steps: 51%|████████████▎ | 5105/10000 [09:35<05:55, 13.76it/s] Update steps: 51%|████████████▎ | 5107/10000 [09:35<05:55, 13.76it/s] Update steps: 51%|████████████▎ | 5109/10000 [09:36<05:55, 13.76it/s] Update steps: 51%|████████████▎ | 5111/10000 [09:36<05:53, 13.84it/s] Update steps: 51%|████████████▎ | 5113/10000 [09:36<05:52, 13.86it/s] Update steps: 51%|████████████▎ | 5115/10000 [09:36<05:53, 13.80it/s] Update steps: 51%|████████████▎ | 5117/10000 [09:36<05:52, 13.85it/s] Update steps: 51%|████████████▎ | 5119/10000 [09:36<05:53, 13.79it/s] Update steps: 51%|████████████▎ | 5121/10000 [09:36<05:52, 13.86it/s] Update steps: 51%|████████████▎ | 5123/10000 [09:37<05:55, 13.71it/s] Update steps: 51%|████████████▎ | 5125/10000 [09:37<05:52, 13.81it/s] Update steps: 51%|████████████▎ | 5127/10000 [09:37<05:50, 13.91it/s] Update steps: 51%|████████████▎ | 5129/10000 [09:37<05:51, 13.88it/s] Update steps: 51%|████████████▎ | 5131/10000 [09:37<05:57, 13.63it/s] Update steps: 51%|████████████▎ | 5133/10000 [09:37<05:53, 13.75it/s] Update steps: 51%|████████████▎ | 5135/10000 [09:37<05:51, 13.82it/s] Update steps: 51%|████████████▎ | 5137/10000 [09:38<05:49, 13.90it/s] Update steps: 51%|████████████▎ | 5139/10000 [09:38<05:52, 13.80it/s] Update steps: 51%|████████████▎ | 5141/10000 [09:38<05:51, 13.81it/s] Update steps: 51%|████████████▎ | 5143/10000 [09:38<05:52, 13.77it/s] Update steps: 51%|████████████▎ | 5145/10000 [09:38<05:52, 13.76it/s] Update steps: 51%|████████████▎ | 5147/10000 [09:38<05:51, 13.82it/s] Update steps: 51%|████████████▎ | 5149/10000 [09:38<05:49, 13.89it/s] Update steps: 52%|████████████▎ | 5151/10000 [09:39<05:46, 13.98it/s] Update steps: 52%|████████████▎ | 5153/10000 [09:39<05:49, 13.86it/s] Update steps: 52%|████████████▎ | 5155/10000 [09:39<05:52, 13.76it/s] Update steps: 52%|████████████▍ | 5157/10000 [09:39<05:52, 13.72it/s] Update steps: 52%|████████████▍ | 5159/10000 [09:39<05:51, 13.78it/s] Update steps: 52%|████████████▍ | 5161/10000 [09:39<05:49, 13.85it/s] Update steps: 52%|████████████▍ | 5163/10000 [09:39<05:50, 13.78it/s] Update steps: 52%|████████████▍ | 5165/10000 [09:40<05:51, 13.76it/s] Update steps: 52%|████████████▍ | 5167/10000 [09:40<05:54, 13.64it/s] Update steps: 52%|████████████▍ | 5169/10000 [09:40<05:52, 13.72it/s] Update steps: 52%|████████████▍ | 5171/10000 [09:40<05:51, 13.76it/s] Update steps: 52%|████████████▍ | 5173/10000 [09:40<05:49, 13.80it/s] Update steps: 52%|████████████▍ | 5175/10000 [09:40<05:51, 13.72it/s] Update steps: 52%|████████████▍ | 5177/10000 [09:40<05:50, 13.75it/s] Update steps: 52%|████████████▍ | 5179/10000 [09:41<05:50, 13.74it/s] Update steps: 52%|████████████▍ | 5181/10000 [09:41<05:53, 13.62it/s] Update steps: 52%|████████████▍ | 5183/10000 [09:41<05:56, 13.51it/s] Update steps: 52%|████████████▍ | 5185/10000 [09:41<05:58, 13.43it/s] Update steps: 52%|████████████▍ | 5187/10000 [09:41<05:53, 13.62it/s] Update steps: 52%|████████████▍ | 5189/10000 [09:41<05:51, 13.67it/s] Update steps: 52%|████████████▍ | 5191/10000 [09:41<05:48, 13.78it/s] Update steps: 52%|████████████▍ | 5193/10000 [09:42<05:47, 13.82it/s] Update steps: 52%|████████████▍ | 5195/10000 [09:42<05:47, 13.83it/s] Update steps: 52%|████████████▍ | 5197/10000 [09:42<05:46, 13.86it/s] Update steps: 52%|████████████▍ | 5199/10000 [09:42<05:45, 13.91it/s] Update steps: 52%|████████████▍ | 5201/10000 [09:42<05:45, 13.88it/s] Update steps: 52%|████████████▍ | 5203/10000 [09:42<05:45, 13.88it/s] Update steps: 52%|████████████▍ | 5205/10000 [09:42<05:52, 13.61it/s] Update steps: 52%|████████████▍ | 5207/10000 [09:43<05:50, 13.67it/s] Update steps: 52%|████████████▌ | 5209/10000 [09:43<05:47, 13.77it/s] Update steps: 52%|████████████▌ | 5211/10000 [09:43<05:46, 13.83it/s] Update steps: 52%|████████████▌ | 5213/10000 [09:43<05:45, 13.84it/s] Update steps: 52%|████████████▌ | 5215/10000 [09:43<05:48, 13.72it/s] Update steps: 52%|████████████▌ | 5217/10000 [09:43<05:50, 13.64it/s] Update steps: 52%|████████████▌ | 5219/10000 [09:44<05:47, 13.77it/s] Update steps: 52%|████████████▌ | 5221/10000 [09:44<05:45, 13.85it/s] Update steps: 52%|████████████▌ | 5223/10000 [09:44<05:45, 13.82it/s] Update steps: 52%|████████████▌ | 5225/10000 [09:44<05:46, 13.77it/s] Update steps: 52%|████████████▌ | 5227/10000 [09:44<05:54, 13.46it/s] Update steps: 52%|████████████▌ | 5229/10000 [09:44<05:54, 13.47it/s] Update steps: 52%|████████████▌ | 5231/10000 [09:44<05:53, 13.47it/s] Update steps: 52%|████████████▌ | 5233/10000 [09:45<05:50, 13.59it/s] Update steps: 52%|████████████▌ | 5235/10000 [09:45<05:47, 13.73it/s] Update steps: 52%|████████████▌ | 5237/10000 [09:45<05:46, 13.73it/s] Update steps: 52%|████████████▌ | 5239/10000 [09:45<05:47, 13.68it/s] Update steps: 52%|████████████▌ | 5241/10000 [09:45<05:47, 13.71it/s] Update steps: 52%|████████████▌ | 5243/10000 [09:45<05:50, 13.56it/s] Update steps: 52%|████████████▌ | 5245/10000 [09:45<05:48, 13.63it/s] Update steps: 52%|████████████▌ | 5247/10000 [09:46<05:50, 13.57it/s] Update steps: 52%|████████████▌ | 5249/10000 [09:46<05:50, 13.54it/s] Update steps: 53%|████████████▌ | 5251/10000 [09:46<05:48, 13.62it/s] Update steps: 53%|████████████▌ | 5253/10000 [09:46<05:45, 13.72it/s] Update steps: 53%|████████████▌ | 5255/10000 [09:46<05:43, 13.82it/s] Update steps: 53%|████████████▌ | 5257/10000 [09:46<05:44, 13.75it/s] Update steps: 53%|████████████▌ | 5259/10000 [09:46<05:44, 13.76it/s] Update steps: 53%|████████████▋ | 5261/10000 [09:47<05:43, 13.80it/s] Update steps: 53%|████████████▋ | 5263/10000 [09:47<05:45, 13.72it/s] Update steps: 53%|████████████▋ | 5265/10000 [09:47<05:46, 13.67it/s] Update steps: 53%|████████████▋ | 5267/10000 [09:47<05:43, 13.78it/s] Update steps: 53%|████████████▋ | 5269/10000 [09:47<05:43, 13.79it/s] Update steps: 53%|████████████▋ | 5271/10000 [09:47<05:41, 13.85it/s] Update steps: 53%|████████████▋ | 5273/10000 [09:47<05:43, 13.78it/s] Update steps: 53%|████████████▋ | 5275/10000 [09:48<05:45, 13.66it/s] Update steps: 53%|████████████▋ | 5277/10000 [09:48<05:45, 13.67it/s] Update steps: 53%|████████████▋ | 5279/10000 [09:48<05:46, 13.64it/s] Update steps: 53%|████████████▋ | 5281/10000 [09:48<05:43, 13.73it/s] Update steps: 53%|████████████▋ | 5283/10000 [09:48<05:43, 13.74it/s] Update steps: 53%|████████████▋ | 5285/10000 [09:48<05:40, 13.83it/s] Update steps: 53%|████████████▋ | 5287/10000 [09:48<05:42, 13.75it/s] Update steps: 53%|████████████▋ | 5289/10000 [09:49<05:41, 13.81it/s] Update steps: 53%|████████████▋ | 5291/10000 [09:49<05:41, 13.81it/s] Update steps: 53%|████████████▋ | 5293/10000 [09:49<05:41, 13.79it/s] Update steps: 53%|████████████▋ | 5295/10000 [09:49<05:42, 13.74it/s] Update steps: 53%|████████████▋ | 5297/10000 [09:49<05:47, 13.53it/s] Update steps: 53%|████████████▋ | 5299/10000 [09:49<05:45, 13.61it/s] Update steps: 53%|████████████▋ | 5301/10000 [09:49<05:44, 13.65it/s] Update steps: 53%|████████████▋ | 5303/10000 [09:50<05:45, 13.60it/s] Update steps: 53%|████████████▋ | 5305/10000 [09:50<05:45, 13.60it/s] Update steps: 53%|████████████▋ | 5307/10000 [09:50<05:44, 13.61it/s] Update steps: 53%|████████████▋ | 5309/10000 [09:50<05:46, 13.56it/s] Update steps: 53%|████████████▋ | 5311/10000 [09:50<05:42, 13.68it/s] Update steps: 53%|████████████▊ | 5313/10000 [09:50<05:42, 13.69it/s] Update steps: 53%|████████████▊ | 5315/10000 [09:51<05:40, 13.77it/s] Update steps: 53%|████████████▊ | 5317/10000 [09:51<05:39, 13.79it/s] Update steps: 53%|████████████▊ | 5319/10000 [09:51<05:38, 13.82it/s] Update steps: 53%|████████████▊ | 5321/10000 [09:51<05:41, 13.71it/s] Update steps: 53%|████████████▊ | 5323/10000 [09:51<05:41, 13.68it/s] Update steps: 53%|████████████▊ | 5325/10000 [09:51<05:41, 13.70it/s] Update steps: 53%|████████████▊ | 5327/10000 [09:51<05:39, 13.76it/s] Update steps: 53%|████████████▊ | 5329/10000 [09:52<05:43, 13.58it/s] Update steps: 53%|████████████▊ | 5331/10000 [09:52<05:43, 13.60it/s] Update steps: 53%|████████████▊ | 5333/10000 [09:52<05:40, 13.70it/s] Update steps: 53%|████████████▊ | 5335/10000 [09:52<05:37, 13.81it/s] Update steps: 53%|████████████▊ | 5337/10000 [09:52<05:39, 13.75it/s] Update steps: 53%|████████████▊ | 5339/10000 [09:52<05:41, 13.66it/s] Update steps: 53%|████████████▊ | 5341/10000 [09:52<05:43, 13.58it/s] Update steps: 53%|████████████▊ | 5343/10000 [09:53<05:39, 13.73it/s] Update steps: 53%|████████████▊ | 5345/10000 [09:53<05:39, 13.70it/s] Update steps: 53%|████████████▊ | 5347/10000 [09:53<05:37, 13.79it/s] Update steps: 53%|████████████▊ | 5349/10000 [09:53<05:37, 13.78it/s] Update steps: 54%|████████████▊ | 5351/10000 [09:53<05:38, 13.74it/s] Update steps: 54%|████████████▊ | 5353/10000 [09:53<05:38, 13.72it/s] Update steps: 54%|████████████▊ | 5355/10000 [09:53<05:43, 13.52it/s] Update steps: 54%|████████████▊ | 5357/10000 [09:54<05:41, 13.61it/s] Update steps: 54%|████████████▊ | 5359/10000 [09:54<05:38, 13.71it/s] Update steps: 54%|████████████▊ | 5361/10000 [09:54<05:40, 13.61it/s] Update steps: 54%|████████████▊ | 5363/10000 [09:54<05:37, 13.73it/s] Update steps: 54%|████████████▉ | 5365/10000 [09:54<05:35, 13.81it/s] Update steps: 54%|████████████▉ | 5367/10000 [09:54<05:36, 13.79it/s] Update steps: 54%|████████████▉ | 5369/10000 [09:54<05:33, 13.88it/s] Update steps: 54%|████████████▉ | 5371/10000 [09:55<05:32, 13.93it/s] Update steps: 54%|████████████▉ | 5373/10000 [09:55<05:32, 13.92it/s] Update steps: 54%|████████████▉ | 5375/10000 [09:55<05:31, 13.96it/s] Update steps: 54%|████████████▉ | 5377/10000 [09:55<05:30, 13.98it/s] Update steps: 54%|████████████▉ | 5379/10000 [09:55<05:32, 13.90it/s] Update steps: 54%|████████████▉ | 5381/10000 [09:55<05:31, 13.94it/s] Update steps: 54%|████████████▉ | 5383/10000 [09:55<05:31, 13.92it/s] Update steps: 54%|████████████▉ | 5385/10000 [09:56<05:32, 13.90it/s] Update steps: 54%|████████████▉ | 5387/10000 [09:56<05:32, 13.88it/s] Update steps: 54%|████████████▉ | 5389/10000 [09:56<05:30, 13.95it/s] Update steps: 54%|████████████▉ | 5391/10000 [09:56<05:31, 13.89it/s] Update steps: 54%|████████████▉ | 5393/10000 [09:56<05:30, 13.93it/s] Update steps: 54%|████████████▉ | 5395/10000 [09:56<05:33, 13.83it/s] Update steps: 54%|████████████▉ | 5397/10000 [09:56<05:31, 13.88it/s] Update steps: 54%|████████████▉ | 5399/10000 [09:57<05:31, 13.90it/s] Update steps: 54%|████████████▉ | 5401/10000 [09:57<05:32, 13.82it/s] Update steps: 54%|████████████▉ | 5403/10000 [09:57<05:36, 13.67it/s] Update steps: 54%|████████████▉ | 5405/10000 [09:57<05:33, 13.78it/s] Update steps: 54%|████████████▉ | 5407/10000 [09:57<05:30, 13.89it/s] Update steps: 54%|████████████▉ | 5409/10000 [09:57<05:29, 13.93it/s] Update steps: 54%|████████████▉ | 5411/10000 [09:57<05:28, 13.96it/s] Update steps: 54%|████████████▉ | 5413/10000 [09:58<05:28, 13.95it/s] Update steps: 54%|████████████▉ | 5415/10000 [09:58<05:27, 13.99it/s] Update steps: 54%|█████████████ | 5417/10000 [09:58<05:29, 13.91it/s] Update steps: 54%|█████████████ | 5419/10000 [09:58<05:30, 13.86it/s] Update steps: 54%|█████████████ | 5421/10000 [09:58<05:28, 13.92it/s] Update steps: 54%|█████████████ | 5423/10000 [09:58<05:29, 13.89it/s] Update steps: 54%|█████████████ | 5425/10000 [09:58<05:31, 13.81it/s] Update steps: 54%|█████████████ | 5427/10000 [09:59<05:36, 13.58it/s] Update steps: 54%|█████████████ | 5429/10000 [09:59<05:34, 13.67it/s] Update steps: 54%|█████████████ | 5431/10000 [09:59<05:31, 13.79it/s] Update steps: 54%|█████████████ | 5433/10000 [09:59<05:28, 13.91it/s] Update steps: 54%|█████████████ | 5435/10000 [09:59<05:27, 13.94it/s] Update steps: 54%|█████████████ | 5437/10000 [09:59<05:27, 13.93it/s] Update steps: 54%|█████████████ | 5439/10000 [10:00<05:33, 13.68it/s] Update steps: 54%|█████████████ | 5441/10000 [10:00<05:32, 13.72it/s] Update steps: 54%|█████████████ | 5443/10000 [10:00<05:29, 13.85it/s] Update steps: 54%|█████████████ | 5445/10000 [10:00<05:27, 13.90it/s] Update steps: 54%|█████████████ | 5447/10000 [10:00<05:29, 13.82it/s] Update steps: 54%|█████████████ | 5449/10000 [10:00<05:27, 13.88it/s] Update steps: 55%|█████████████ | 5451/10000 [10:00<05:28, 13.84it/s] Update steps: 55%|█████████████ | 5453/10000 [10:01<05:27, 13.89it/s] Update steps: 55%|█████████████ | 5455/10000 [10:01<05:27, 13.89it/s] Update steps: 55%|█████████████ | 5457/10000 [10:01<05:29, 13.80it/s] Update steps: 55%|█████████████ | 5459/10000 [10:01<05:26, 13.89it/s] Update steps: 55%|█████████████ | 5461/10000 [10:01<05:30, 13.75it/s] Update steps: 55%|█████████████ | 5463/10000 [10:01<05:28, 13.82it/s] Update steps: 55%|█████████████ | 5465/10000 [10:01<05:26, 13.91it/s] Update steps: 55%|█████████████ | 5467/10000 [10:02<05:27, 13.84it/s] Update steps: 55%|█████████████▏ | 5469/10000 [10:02<05:26, 13.88it/s] Update steps: 55%|█████████████▏ | 5471/10000 [10:02<05:26, 13.89it/s] Update steps: 55%|█████████████▏ | 5473/10000 [10:02<05:24, 13.95it/s] Update steps: 55%|█████████████▏ | 5475/10000 [10:02<05:23, 13.98it/s] Update steps: 55%|█████████████▏ | 5477/10000 [10:02<05:23, 13.98it/s] Update steps: 55%|█████████████▏ | 5479/10000 [10:02<05:23, 13.97it/s] Update steps: 55%|█████████████▏ | 5481/10000 [10:03<05:22, 14.01it/s] Update steps: 55%|█████████████▏ | 5483/10000 [10:03<05:25, 13.89it/s] Update steps: 55%|█████████████▏ | 5485/10000 [10:03<05:27, 13.78it/s] Update steps: 55%|█████████████▏ | 5487/10000 [10:03<05:26, 13.83it/s] Update steps: 55%|█████████████▏ | 5489/10000 [10:03<05:24, 13.88it/s] Update steps: 55%|█████████████▏ | 5491/10000 [10:03<05:23, 13.93it/s] Update steps: 55%|█████████████▏ | 5493/10000 [10:03<05:25, 13.84it/s] Update steps: 55%|█████████████▏ | 5495/10000 [10:04<05:24, 13.90it/s] Update steps: 55%|█████████████▏ | 5497/10000 [10:04<05:23, 13.92it/s] Update steps: 55%|█████████████▏ | 5499/10000 [10:04<05:25, 13.84it/s] Update steps: 55%|█████████████▏ | 5501/10000 [10:04<05:23, 13.89it/s] Update steps: 55%|█████████████▏ | 5503/10000 [10:04<05:22, 13.92it/s] Update steps: 55%|█████████████▏ | 5505/10000 [10:04<05:24, 13.86it/s] Update steps: 55%|█████████████▏ | 5507/10000 [10:04<05:22, 13.93it/s] Update steps: 55%|█████████████▏ | 5509/10000 [10:05<05:26, 13.77it/s] Update steps: 55%|█████████████▏ | 5511/10000 [10:05<05:26, 13.75it/s] Update steps: 55%|█████████████▏ | 5513/10000 [10:05<05:24, 13.83it/s] Update steps: 55%|█████████████▏ | 5515/10000 [10:05<05:24, 13.81it/s] Update steps: 55%|█████████████▏ | 5517/10000 [10:05<05:24, 13.80it/s] Update steps: 55%|█████████████▏ | 5519/10000 [10:05<05:22, 13.88it/s] Update steps: 55%|█████████████▎ | 5521/10000 [10:05<05:22, 13.87it/s] Update steps: 55%|█████████████▎ | 5523/10000 [10:06<05:22, 13.89it/s] Update steps: 55%|█████████████▎ | 5525/10000 [10:06<05:21, 13.92it/s] Update steps: 55%|█████████████▎ | 5527/10000 [10:06<05:20, 13.94it/s] Update steps: 55%|█████████████▎ | 5529/10000 [10:06<05:20, 13.97it/s] Update steps: 55%|█████████████▎ | 5531/10000 [10:06<05:20, 13.95it/s] Update steps: 55%|█████████████▎ | 5533/10000 [10:06<05:19, 13.97it/s] Update steps: 55%|█████████████▎ | 5535/10000 [10:06<05:22, 13.86it/s] Update steps: 55%|█████████████▎ | 5537/10000 [10:07<05:22, 13.85it/s] Update steps: 55%|█████████████▎ | 5539/10000 [10:07<05:24, 13.75it/s] Update steps: 55%|█████████████▎ | 5541/10000 [10:07<05:23, 13.79it/s] Update steps: 55%|█████████████▎ | 5543/10000 [10:07<05:21, 13.88it/s] Update steps: 55%|█████████████▎ | 5545/10000 [10:07<05:21, 13.86it/s] Update steps: 55%|█████████████▎ | 5547/10000 [10:07<05:25, 13.70it/s] Update steps: 55%|█████████████▎ | 5549/10000 [10:07<05:22, 13.81it/s] Update steps: 56%|█████████████▎ | 5551/10000 [10:08<05:20, 13.87it/s] Update steps: 56%|█████████████▎ | 5553/10000 [10:08<05:19, 13.91it/s] Update steps: 56%|█████████████▎ | 5555/10000 [10:08<05:20, 13.88it/s] Update steps: 56%|█████████████▎ | 5557/10000 [10:08<05:19, 13.91it/s] Update steps: 56%|█████████████▎ | 5559/10000 [10:08<05:19, 13.90it/s] Update steps: 56%|█████████████▎ | 5561/10000 [10:08<05:19, 13.88it/s] Update steps: 56%|█████████████▎ | 5563/10000 [10:08<05:21, 13.80it/s] Update steps: 56%|█████████████▎ | 5565/10000 [10:09<05:21, 13.79it/s] Update steps: 56%|█████████████▎ | 5567/10000 [10:09<05:21, 13.81it/s] Update steps: 56%|█████████████▎ | 5569/10000 [10:09<05:20, 13.81it/s] Update steps: 56%|█████████████▎ | 5571/10000 [10:09<05:20, 13.81it/s] Update steps: 56%|█████████████▍ | 5573/10000 [10:09<05:18, 13.88it/s] Update steps: 56%|█████████████▍ | 5575/10000 [10:09<05:16, 13.96it/s] Update steps: 56%|█████████████▍ | 5577/10000 [10:09<05:15, 14.01it/s] Update steps: 56%|█████████████▍ | 5579/10000 [10:10<05:15, 14.01it/s] Update steps: 56%|█████████████▍ | 5581/10000 [10:10<05:14, 14.07it/s] Update steps: 56%|█████████████▍ | 5583/10000 [10:10<05:13, 14.09it/s] Update steps: 56%|█████████████▍ | 5585/10000 [10:10<05:16, 13.96it/s] Update steps: 56%|█████████████▍ | 5587/10000 [10:10<05:16, 13.96it/s] Update steps: 56%|█████████████▍ | 5589/10000 [10:10<05:15, 14.00it/s] Update steps: 56%|█████████████▍ | 5591/10000 [10:10<05:17, 13.89it/s] Update steps: 56%|█████████████▍ | 5593/10000 [10:11<05:15, 13.97it/s] Update steps: 56%|█████████████▍ | 5595/10000 [10:11<05:17, 13.90it/s] Update steps: 56%|█████████████▍ | 5597/10000 [10:11<05:19, 13.79it/s] Update steps: 56%|█████████████▍ | 5599/10000 [10:11<05:17, 13.87it/s] Update steps: 56%|█████████████▍ | 5601/10000 [10:11<05:16, 13.89it/s] Update steps: 56%|█████████████▍ | 5603/10000 [10:11<05:15, 13.94it/s] Update steps: 56%|█████████████▍ | 5605/10000 [10:11<05:17, 13.83it/s] Update steps: 56%|█████████████▍ | 5607/10000 [10:12<05:16, 13.87it/s] Update steps: 56%|█████████████▍ | 5609/10000 [10:12<05:20, 13.71it/s] Update steps: 56%|█████████████▍ | 5611/10000 [10:12<05:19, 13.74it/s] Update steps: 56%|█████████████▍ | 5613/10000 [10:12<05:16, 13.85it/s] Update steps: 56%|█████████████▍ | 5615/10000 [10:12<05:18, 13.78it/s] Update steps: 56%|█████████████▍ | 5617/10000 [10:12<05:17, 13.83it/s] Update steps: 56%|█████████████▍ | 5619/10000 [10:12<05:15, 13.88it/s] Update steps: 56%|█████████████▍ | 5621/10000 [10:13<05:16, 13.83it/s] Update steps: 56%|█████████████▍ | 5623/10000 [10:13<05:19, 13.69it/s] Update steps: 56%|█████████████▌ | 5625/10000 [10:13<05:17, 13.79it/s] Update steps: 56%|█████████████▌ | 5627/10000 [10:13<05:15, 13.85it/s] Update steps: 56%|█████████████▌ | 5629/10000 [10:13<05:13, 13.95it/s] Update steps: 56%|█████████████▌ | 5631/10000 [10:13<05:12, 13.99it/s] Update steps: 56%|█████████████▌ | 5633/10000 [10:13<05:11, 14.00it/s] Update steps: 56%|█████████████▌ | 5635/10000 [10:14<05:15, 13.85it/s] Update steps: 56%|█████████████▌ | 5637/10000 [10:14<05:17, 13.72it/s] Update steps: 56%|█████████████▌ | 5639/10000 [10:14<05:18, 13.68it/s] Update steps: 56%|█████████████▌ | 5641/10000 [10:14<05:17, 13.75it/s] Update steps: 56%|█████████████▌ | 5643/10000 [10:14<05:16, 13.77it/s] Update steps: 56%|█████████████▌ | 5645/10000 [10:14<05:14, 13.84it/s] Update steps: 56%|█████████████▌ | 5647/10000 [10:15<05:13, 13.86it/s] Update steps: 56%|█████████████▌ | 5649/10000 [10:15<05:12, 13.92it/s] Update steps: 57%|█████████████▌ | 5651/10000 [10:15<05:14, 13.85it/s] Update steps: 57%|█████████████▌ | 5653/10000 [10:15<05:13, 13.85it/s] Update steps: 57%|█████████████▌ | 5655/10000 [10:15<05:11, 13.93it/s] Update steps: 57%|█████████████▌ | 5657/10000 [10:15<05:10, 13.97it/s] Update steps: 57%|█████████████▌ | 5659/10000 [10:15<05:10, 13.96it/s] Update steps: 57%|█████████████▌ | 5661/10000 [10:16<05:12, 13.89it/s] Update steps: 57%|█████████████▌ | 5663/10000 [10:16<05:11, 13.94it/s] Update steps: 57%|█████████████▌ | 5665/10000 [10:16<05:10, 13.97it/s] Update steps: 57%|█████████████▌ | 5667/10000 [10:16<05:10, 13.96it/s] Update steps: 57%|█████████████▌ | 5669/10000 [10:16<05:14, 13.75it/s] Update steps: 57%|█████████████▌ | 5671/10000 [10:16<05:12, 13.84it/s] Update steps: 57%|█████████████▌ | 5673/10000 [10:16<05:12, 13.85it/s] Update steps: 57%|█████████████▌ | 5675/10000 [10:17<05:11, 13.88it/s] Update steps: 57%|█████████████▌ | 5677/10000 [10:17<05:11, 13.88it/s] Update steps: 57%|█████████████▋ | 5679/10000 [10:17<05:10, 13.91it/s] Update steps: 57%|█████████████▋ | 5681/10000 [10:17<05:10, 13.90it/s] Update steps: 57%|█████████████▋ | 5683/10000 [10:17<05:09, 13.97it/s] Update steps: 57%|█████████████▋ | 5685/10000 [10:17<05:13, 13.78it/s] Update steps: 57%|█████████████▋ | 5687/10000 [10:17<05:12, 13.82it/s] Update steps: 57%|█████████████▋ | 5689/10000 [10:18<05:11, 13.85it/s] Update steps: 57%|█████████████▋ | 5691/10000 [10:18<05:10, 13.90it/s] Update steps: 57%|█████████████▋ | 5693/10000 [10:18<05:10, 13.87it/s] Update steps: 57%|█████████████▋ | 5695/10000 [10:18<05:09, 13.90it/s] Update steps: 57%|█████████████▋ | 5697/10000 [10:18<05:08, 13.96it/s] Update steps: 57%|█████████████▋ | 5699/10000 [10:18<05:12, 13.77it/s] Update steps: 57%|█████████████▋ | 5701/10000 [10:18<05:10, 13.85it/s] Update steps: 57%|█████████████▋ | 5703/10000 [10:19<05:09, 13.89it/s] Update steps: 57%|█████████████▋ | 5705/10000 [10:19<05:08, 13.92it/s] Update steps: 57%|█████████████▋ | 5707/10000 [10:19<05:08, 13.90it/s] Update steps: 57%|█████████████▋ | 5709/10000 [10:19<05:10, 13.84it/s] Update steps: 57%|█████████████▋ | 5711/10000 [10:19<05:08, 13.89it/s] Update steps: 57%|█████████████▋ | 5713/10000 [10:19<05:09, 13.86it/s] Update steps: 57%|█████████████▋ | 5715/10000 [10:19<05:08, 13.89it/s] Update steps: 57%|█████████████▋ | 5717/10000 [10:20<05:08, 13.88it/s] Update steps: 57%|█████████████▋ | 5719/10000 [10:20<05:08, 13.89it/s] Update steps: 57%|█████████████▋ | 5721/10000 [10:20<05:07, 13.91it/s] Update steps: 57%|█████████████▋ | 5723/10000 [10:20<05:13, 13.63it/s] Update steps: 57%|█████████████▋ | 5725/10000 [10:20<05:12, 13.67it/s] Update steps: 57%|█████████████▋ | 5727/10000 [10:20<05:10, 13.75it/s] Update steps: 57%|█████████████▋ | 5729/10000 [10:20<05:09, 13.79it/s] Update steps: 57%|█████████████▊ | 5731/10000 [10:21<05:10, 13.75it/s] Update steps: 57%|█████████████▊ | 5733/10000 [10:21<05:10, 13.76it/s] Update steps: 57%|█████████████▊ | 5735/10000 [10:21<05:09, 13.78it/s] Update steps: 57%|█████████████▊ | 5737/10000 [10:21<05:10, 13.75it/s] Update steps: 57%|█████████████▊ | 5739/10000 [10:21<05:09, 13.77it/s] Update steps: 57%|█████████████▊ | 5741/10000 [10:21<05:09, 13.76it/s] Update steps: 57%|█████████████▊ | 5743/10000 [10:21<05:09, 13.77it/s] Update steps: 57%|█████████████▊ | 5745/10000 [10:22<05:09, 13.76it/s] Update steps: 57%|█████████████▊ | 5747/10000 [10:22<05:07, 13.81it/s] Update steps: 57%|█████████████▊ | 5749/10000 [10:22<05:06, 13.88it/s] Update steps: 58%|█████████████▊ | 5751/10000 [10:22<05:07, 13.82it/s] Update steps: 58%|█████████████▊ | 5753/10000 [10:22<05:07, 13.80it/s] Update steps: 58%|█████████████▊ | 5755/10000 [10:22<05:06, 13.87it/s] Update steps: 58%|█████████████▊ | 5757/10000 [10:22<05:06, 13.86it/s] Update steps: 58%|█████████████▊ | 5759/10000 [10:23<05:05, 13.88it/s] Update steps: 58%|█████████████▊ | 5761/10000 [10:23<05:04, 13.94it/s] Update steps: 58%|█████████████▊ | 5763/10000 [10:23<05:03, 13.94it/s] Update steps: 58%|█████████████▊ | 5765/10000 [10:23<05:03, 13.96it/s] Update steps: 58%|█████████████▊ | 5767/10000 [10:23<05:04, 13.89it/s] Update steps: 58%|█████████████▊ | 5769/10000 [10:23<05:03, 13.94it/s] Update steps: 58%|█████████████▊ | 5771/10000 [10:23<05:04, 13.87it/s] Update steps: 58%|█████████████▊ | 5773/10000 [10:24<05:08, 13.70it/s] Update steps: 58%|█████████████▊ | 5775/10000 [10:24<05:09, 13.66it/s] Update steps: 58%|█████████████▊ | 5777/10000 [10:24<05:06, 13.76it/s] Update steps: 58%|█████████████▊ | 5779/10000 [10:24<05:08, 13.67it/s] Update steps: 58%|█████████████▊ | 5781/10000 [10:24<05:06, 13.77it/s] Update steps: 58%|█████████████▉ | 5783/10000 [10:24<05:10, 13.59it/s] Update steps: 58%|█████████████▉ | 5785/10000 [10:24<05:10, 13.57it/s] Update steps: 58%|█████████████▉ | 5787/10000 [10:25<05:09, 13.61it/s] Update steps: 58%|█████████████▉ | 5789/10000 [10:25<05:08, 13.66it/s] Update steps: 58%|█████████████▉ | 5791/10000 [10:25<05:07, 13.69it/s] Update steps: 58%|█████████████▉ | 5793/10000 [10:25<05:05, 13.79it/s] Update steps: 58%|█████████████▉ | 5795/10000 [10:25<05:03, 13.87it/s] Update steps: 58%|█████████████▉ | 5797/10000 [10:25<05:05, 13.74it/s] Update steps: 58%|█████████████▉ | 5799/10000 [10:25<05:03, 13.86it/s] Update steps: 58%|█████████████▉ | 5801/10000 [10:26<05:01, 13.92it/s] Update steps: 58%|█████████████▉ | 5803/10000 [10:26<05:01, 13.92it/s] Update steps: 58%|█████████████▉ | 5805/10000 [10:26<05:00, 13.98it/s] Update steps: 58%|█████████████▉ | 5807/10000 [10:26<05:00, 13.97it/s] Update steps: 58%|█████████████▉ | 5809/10000 [10:26<04:59, 14.00it/s] Update steps: 58%|█████████████▉ | 5811/10000 [10:26<04:59, 14.00it/s] Update steps: 58%|█████████████▉ | 5813/10000 [10:26<04:59, 14.00it/s] Update steps: 58%|█████████████▉ | 5815/10000 [10:27<04:58, 14.01it/s] Update steps: 58%|█████████████▉ | 5817/10000 [10:27<05:02, 13.84it/s] Update steps: 58%|█████████████▉ | 5819/10000 [10:27<05:01, 13.88it/s] Update steps: 58%|█████████████▉ | 5821/10000 [10:27<05:00, 13.88it/s] Update steps: 58%|█████████████▉ | 5823/10000 [10:27<05:01, 13.84it/s] Update steps: 58%|█████████████▉ | 5825/10000 [10:27<05:03, 13.74it/s] Update steps: 58%|█████████████▉ | 5827/10000 [10:28<05:04, 13.71it/s] Update steps: 58%|█████████████▉ | 5829/10000 [10:28<05:03, 13.73it/s] Update steps: 58%|█████████████▉ | 5831/10000 [10:28<05:04, 13.71it/s] Update steps: 58%|█████████████▉ | 5833/10000 [10:28<05:01, 13.84it/s] Update steps: 58%|██████████████ | 5835/10000 [10:28<04:59, 13.91it/s] Update steps: 58%|██████████████ | 5837/10000 [10:28<05:01, 13.80it/s] Update steps: 58%|██████████████ | 5839/10000 [10:28<05:02, 13.76it/s] Update steps: 58%|██████████████ | 5841/10000 [10:29<05:01, 13.78it/s] Update steps: 58%|██████████████ | 5843/10000 [10:29<05:01, 13.77it/s] Update steps: 58%|██████████████ | 5845/10000 [10:29<05:02, 13.75it/s] Update steps: 58%|██████████████ | 5847/10000 [10:29<05:00, 13.84it/s] Update steps: 58%|██████████████ | 5849/10000 [10:29<04:58, 13.91it/s] Update steps: 59%|██████████████ | 5851/10000 [10:29<05:05, 13.58it/s] Update steps: 59%|██████████████ | 5853/10000 [10:29<05:02, 13.73it/s] Update steps: 59%|██████████████ | 5855/10000 [10:30<05:00, 13.78it/s] Update steps: 59%|██████████████ | 5857/10000 [10:30<05:02, 13.69it/s] Update steps: 59%|██████████████ | 5859/10000 [10:30<05:03, 13.64it/s] Update steps: 59%|██████████████ | 5861/10000 [10:30<05:02, 13.68it/s] Update steps: 59%|██████████████ | 5863/10000 [10:30<05:00, 13.76it/s] Update steps: 59%|██████████████ | 5865/10000 [10:30<04:58, 13.85it/s] Update steps: 59%|██████████████ | 5867/10000 [10:30<04:57, 13.88it/s] Update steps: 59%|██████████████ | 5869/10000 [10:31<04:57, 13.87it/s] Update steps: 59%|██████████████ | 5871/10000 [10:31<04:56, 13.91it/s] Update steps: 59%|██████████████ | 5873/10000 [10:31<05:00, 13.75it/s] Update steps: 59%|██████████████ | 5875/10000 [10:31<04:58, 13.80it/s] Update steps: 59%|██████████████ | 5877/10000 [10:31<04:58, 13.80it/s] Update steps: 59%|██████████████ | 5879/10000 [10:31<05:00, 13.71it/s] Update steps: 59%|██████████████ | 5881/10000 [10:31<04:58, 13.82it/s] Update steps: 59%|██████████████ | 5883/10000 [10:32<04:56, 13.86it/s] Update steps: 59%|██████████████ | 5885/10000 [10:32<04:55, 13.91it/s] Update steps: 59%|██████████████▏ | 5887/10000 [10:32<04:54, 13.98it/s] Update steps: 59%|██████████████▏ | 5889/10000 [10:32<04:54, 13.95it/s] Update steps: 59%|██████████████▏ | 5891/10000 [10:32<04:54, 13.94it/s] Update steps: 59%|██████████████▏ | 5893/10000 [10:32<04:54, 13.94it/s] Update steps: 59%|██████████████▏ | 5895/10000 [10:32<04:54, 13.95it/s] Update steps: 59%|██████████████▏ | 5897/10000 [10:33<04:53, 13.98it/s] Update steps: 59%|██████████████▏ | 5899/10000 [10:33<04:52, 14.01it/s] Update steps: 59%|██████████████▏ | 5901/10000 [10:33<04:54, 13.94it/s] Update steps: 59%|██████████████▏ | 5903/10000 [10:33<04:55, 13.88it/s] Update steps: 59%|██████████████▏ | 5905/10000 [10:33<04:57, 13.76it/s] Update steps: 59%|██████████████▏ | 5907/10000 [10:33<04:56, 13.80it/s] Update steps: 59%|██████████████▏ | 5909/10000 [10:33<04:56, 13.80it/s] Update steps: 59%|██████████████▏ | 5911/10000 [10:34<04:54, 13.86it/s] Update steps: 59%|██████████████▏ | 5913/10000 [10:34<04:53, 13.93it/s] Update steps: 59%|██████████████▏ | 5915/10000 [10:34<04:54, 13.87it/s] Update steps: 59%|██████████████▏ | 5917/10000 [10:34<04:52, 13.94it/s] Update steps: 59%|██████████████▏ | 5919/10000 [10:34<04:53, 13.91it/s] Update steps: 59%|██████████████▏ | 5921/10000 [10:34<04:52, 13.93it/s] Update steps: 59%|██████████████▏ | 5923/10000 [10:34<04:53, 13.91it/s] Update steps: 59%|██████████████▏ | 5925/10000 [10:35<04:52, 13.95it/s] Update steps: 59%|██████████████▏ | 5927/10000 [10:35<04:53, 13.87it/s] Update steps: 59%|██████████████▏ | 5929/10000 [10:35<04:53, 13.86it/s] Update steps: 59%|██████████████▏ | 5931/10000 [10:35<04:54, 13.83it/s] Update steps: 59%|██████████████▏ | 5933/10000 [10:35<04:53, 13.84it/s] Update steps: 59%|██████████████▏ | 5935/10000 [10:35<04:52, 13.91it/s] Update steps: 59%|██████████████▏ | 5937/10000 [10:35<04:50, 13.97it/s] Update steps: 59%|██████████████▎ | 5939/10000 [10:36<04:50, 14.00it/s] Update steps: 59%|██████████████▎ | 5941/10000 [10:36<04:48, 14.05it/s] Update steps: 59%|██████████████▎ | 5943/10000 [10:36<04:48, 14.08it/s] Update steps: 59%|██████████████▎ | 5945/10000 [10:36<04:48, 14.07it/s] Update steps: 59%|██████████████▎ | 5947/10000 [10:36<04:47, 14.08it/s] Update steps: 59%|██████████████▎ | 5949/10000 [10:36<04:49, 13.99it/s] Update steps: 60%|██████████████▎ | 5951/10000 [10:36<04:48, 14.01it/s] Update steps: 60%|██████████████▎ | 5953/10000 [10:37<04:49, 14.00it/s] Update steps: 60%|██████████████▎ | 5955/10000 [10:37<04:49, 13.99it/s] Update steps: 60%|██████████████▎ | 5957/10000 [10:37<04:50, 13.93it/s] Update steps: 60%|██████████████▎ | 5959/10000 [10:37<04:50, 13.89it/s] Update steps: 60%|██████████████▎ | 5961/10000 [10:37<04:50, 13.90it/s] Update steps: 60%|██████████████▎ | 5963/10000 [10:37<04:49, 13.96it/s] Update steps: 60%|██████████████▎ | 5965/10000 [10:37<04:49, 13.95it/s] Update steps: 60%|██████████████▎ | 5967/10000 [10:38<04:56, 13.62it/s] Update steps: 60%|██████████████▎ | 5969/10000 [10:38<04:55, 13.66it/s] Update steps: 60%|██████████████▎ | 5971/10000 [10:38<04:54, 13.69it/s] Update steps: 60%|██████████████▎ | 5973/10000 [10:38<04:55, 13.62it/s] Update steps: 60%|██████████████▎ | 5975/10000 [10:38<04:56, 13.56it/s] Update steps: 60%|██████████████▎ | 5977/10000 [10:38<04:56, 13.56it/s] Update steps: 60%|██████████████▎ | 5979/10000 [10:38<04:54, 13.65it/s] Update steps: 60%|██████████████▎ | 5981/10000 [10:39<04:54, 13.65it/s] Update steps: 60%|██████████████▎ | 5983/10000 [10:39<04:58, 13.45it/s] Update steps: 60%|██████████████▎ | 5985/10000 [10:39<04:53, 13.66it/s] Update steps: 60%|██████████████▎ | 5987/10000 [10:39<04:54, 13.65it/s] Update steps: 60%|██████████████▎ | 5989/10000 [10:39<04:50, 13.79it/s] Update steps: 60%|██████████████▍ | 5991/10000 [10:39<04:50, 13.79it/s] Update steps: 60%|██████████████▍ | 5993/10000 [10:40<04:53, 13.66it/s] Update steps: 60%|██████████████▍ | 5995/10000 [10:40<04:50, 13.77it/s] Update steps: 60%|██████████████▍ | 5997/10000 [10:40<04:50, 13.77it/s] Update steps: 60%|██████████████▍ | 5999/10000 [10:40<04:50, 13.75it/s]2026-04-17 11:04:00.905 | INFO | __main__:main:490 - Performing evaluation at step 6000 +2026-04-17 11:04:00.907 | INFO | utils.eval:evaluate_model:21 - Loading validation data from 8 local validation files +2026-04-17 11:04:01.661 | INFO | utils.eval:evaluate_model:43 - Loaded validation dataset in 0.76 seconds +2026-04-17 11:04:01.661 | INFO | utils.eval:evaluate_model:60 - Eval set prepared in 0.76 seconds + Update steps: 60%|██████████████▍ | 6000/10000 [10:56<04:50, 13.75it/s]2026-04-17 11:04:37.550 | INFO | __main__:main:505 - Eval loss at step 6000: 3.5013622117901217, Eval perplexity: 33.16059295660506 + Update steps: 60%|█████████████▏ | 6001/10000 [11:17<6:11:11, 5.57s/it] Update steps: 60%|█████████████▏ | 6003/10000 [11:17<4:21:06, 3.92s/it] Update steps: 60%|█████████████▏ | 6005/10000 [11:17<3:04:07, 2.77s/it] Update steps: 60%|█████████████▏ | 6007/10000 [11:17<2:10:16, 1.96s/it] Update steps: 60%|█████████████▏ | 6009/10000 [11:17<1:32:36, 1.39s/it] Update steps: 60%|█████████████▏ | 6011/10000 [11:17<1:06:15, 1.00it/s] Update steps: 60%|██████████████▍ | 6013/10000 [11:18<47:48, 1.39it/s] Update steps: 60%|██████████████▍ | 6015/10000 [11:18<34:53, 1.90it/s] Update steps: 60%|██████████████▍ | 6017/10000 [11:18<25:51, 2.57it/s] Update steps: 60%|██████████████▍ | 6019/10000 [11:18<19:30, 3.40it/s] Update steps: 60%|██████████████▍ | 6021/10000 [11:18<15:03, 4.40it/s] Update steps: 60%|██████████████▍ | 6023/10000 [11:18<12:00, 5.52it/s] Update steps: 60%|██████████████▍ | 6025/10000 [11:18<09:48, 6.76it/s] Update steps: 60%|██████████████▍ | 6027/10000 [11:19<08:16, 8.01it/s] Update steps: 60%|██████████████▍ | 6029/10000 [11:19<07:13, 9.16it/s] Update steps: 60%|██████████████▍ | 6031/10000 [11:19<06:28, 10.22it/s] Update steps: 60%|██████████████▍ | 6033/10000 [11:19<05:57, 11.09it/s] Update steps: 60%|██████████████▍ | 6035/10000 [11:19<05:34, 11.86it/s] Update steps: 60%|██████████████▍ | 6037/10000 [11:19<05:20, 12.35it/s] Update steps: 60%|██████████████▍ | 6039/10000 [11:19<05:11, 12.74it/s] Update steps: 60%|██████████████▍ | 6041/10000 [11:20<05:02, 13.09it/s] Update steps: 60%|██████████████▌ | 6043/10000 [11:20<04:56, 13.36it/s] Update steps: 60%|██████████████▌ | 6045/10000 [11:20<04:53, 13.45it/s] Update steps: 60%|██████████████▌ | 6047/10000 [11:20<04:50, 13.59it/s] Update steps: 60%|██████████████▌ | 6049/10000 [11:20<04:48, 13.71it/s] Update steps: 61%|██████████████▌ | 6051/10000 [11:20<04:49, 13.65it/s] Update steps: 61%|██████████████▌ | 6053/10000 [11:20<04:50, 13.59it/s] Update steps: 61%|██████████████▌ | 6055/10000 [11:21<04:48, 13.66it/s] Update steps: 61%|██████████████▌ | 6057/10000 [11:21<04:49, 13.63it/s] Update steps: 61%|██████████████▌ | 6059/10000 [11:21<04:46, 13.75it/s] Update steps: 61%|██████████████▌ | 6061/10000 [11:21<04:49, 13.62it/s] Update steps: 61%|██████████████▌ | 6063/10000 [11:21<04:50, 13.56it/s] Update steps: 61%|██████████████▌ | 6065/10000 [11:21<04:52, 13.48it/s] Update steps: 61%|██████████████▌ | 6067/10000 [11:22<04:50, 13.53it/s] Update steps: 61%|██████████████▌ | 6069/10000 [11:22<04:50, 13.52it/s] Update steps: 61%|██████████████▌ | 6071/10000 [11:22<04:51, 13.46it/s] Update steps: 61%|██████████████▌ | 6073/10000 [11:22<04:47, 13.66it/s] Update steps: 61%|██████████████▌ | 6075/10000 [11:22<04:48, 13.62it/s] Update steps: 61%|██████████████▌ | 6077/10000 [11:22<04:48, 13.60it/s] Update steps: 61%|██████████████▌ | 6079/10000 [11:22<04:46, 13.68it/s] Update steps: 61%|██████████████▌ | 6081/10000 [11:23<04:45, 13.75it/s] Update steps: 61%|██████████████▌ | 6083/10000 [11:23<04:43, 13.82it/s] Update steps: 61%|██████████████▌ | 6085/10000 [11:23<04:43, 13.80it/s] Update steps: 61%|██████████████▌ | 6087/10000 [11:23<04:45, 13.69it/s] Update steps: 61%|██████████████▌ | 6089/10000 [11:23<04:46, 13.66it/s] Update steps: 61%|██████████████▌ | 6091/10000 [11:23<04:48, 13.57it/s] Update steps: 61%|██████████████▌ | 6093/10000 [11:23<04:46, 13.61it/s] Update steps: 61%|██████████████▋ | 6095/10000 [11:24<04:50, 13.46it/s] Update steps: 61%|██████████████▋ | 6097/10000 [11:24<04:46, 13.62it/s] Update steps: 61%|██████████████▋ | 6099/10000 [11:24<04:43, 13.75it/s] Update steps: 61%|██████████████▋ | 6101/10000 [11:24<04:42, 13.80it/s] Update steps: 61%|██████████████▋ | 6103/10000 [11:24<04:41, 13.84it/s] Update steps: 61%|██████████████▋ | 6105/10000 [11:24<04:40, 13.87it/s] Update steps: 61%|██████████████▋ | 6107/10000 [11:24<04:39, 13.94it/s] Update steps: 61%|██████████████▋ | 6109/10000 [11:25<04:39, 13.94it/s] Update steps: 61%|██████████████▋ | 6111/10000 [11:25<04:40, 13.87it/s] Update steps: 61%|██████████████▋ | 6113/10000 [11:25<04:40, 13.87it/s] Update steps: 61%|██████████████▋ | 6115/10000 [11:25<04:40, 13.87it/s] Update steps: 61%|██████████████▋ | 6117/10000 [11:25<04:41, 13.80it/s] Update steps: 61%|██████████████▋ | 6119/10000 [11:25<04:44, 13.63it/s] Update steps: 61%|██████████████▋ | 6121/10000 [11:25<04:42, 13.72it/s] Update steps: 61%|██████████████▋ | 6123/10000 [11:26<04:40, 13.84it/s] Update steps: 61%|██████████████▋ | 6125/10000 [11:26<04:39, 13.87it/s] Update steps: 61%|██████████████▋ | 6127/10000 [11:26<04:38, 13.89it/s] Update steps: 61%|██████████████▋ | 6129/10000 [11:26<04:37, 13.95it/s] Update steps: 61%|██████████████▋ | 6131/10000 [11:26<04:38, 13.91it/s] Update steps: 61%|██████████████▋ | 6133/10000 [11:26<04:46, 13.50it/s] Update steps: 61%|██████████████▋ | 6135/10000 [11:26<04:44, 13.57it/s] Update steps: 61%|██████████████▋ | 6137/10000 [11:27<04:41, 13.70it/s] Update steps: 61%|██████████████▋ | 6139/10000 [11:27<04:40, 13.77it/s] Update steps: 61%|██████████████▋ | 6141/10000 [11:27<04:38, 13.85it/s] Update steps: 61%|██████████████▋ | 6143/10000 [11:27<04:37, 13.90it/s] Update steps: 61%|██████████████▋ | 6145/10000 [11:27<04:36, 13.95it/s] Update steps: 61%|██████████████▊ | 6147/10000 [11:27<04:36, 13.93it/s] Update steps: 61%|██████████████▊ | 6149/10000 [11:27<04:38, 13.85it/s] Update steps: 62%|██████████████▊ | 6151/10000 [11:28<04:37, 13.85it/s] Update steps: 62%|██████████████▊ | 6153/10000 [11:28<04:38, 13.82it/s] Update steps: 62%|██████████████▊ | 6155/10000 [11:28<04:38, 13.83it/s] Update steps: 62%|██████████████▊ | 6157/10000 [11:28<04:36, 13.90it/s] Update steps: 62%|██████████████▊ | 6159/10000 [11:28<04:37, 13.86it/s] Update steps: 62%|██████████████▊ | 6161/10000 [11:28<04:41, 13.62it/s] Update steps: 62%|██████████████▊ | 6163/10000 [11:28<04:41, 13.65it/s] Update steps: 62%|██████████████▊ | 6165/10000 [11:29<04:38, 13.76it/s] Update steps: 62%|██████████████▊ | 6167/10000 [11:29<04:39, 13.71it/s] Update steps: 62%|██████████████▊ | 6169/10000 [11:29<04:37, 13.80it/s] Update steps: 62%|██████████████▊ | 6171/10000 [11:29<04:40, 13.63it/s] Update steps: 62%|██████████████▊ | 6173/10000 [11:29<04:39, 13.70it/s] Update steps: 62%|██████████████▊ | 6175/10000 [11:29<04:39, 13.69it/s] Update steps: 62%|██████████████▊ | 6177/10000 [11:30<04:39, 13.66it/s] Update steps: 62%|██████████████▊ | 6179/10000 [11:30<04:46, 13.32it/s] Update steps: 62%|██████████████▊ | 6181/10000 [11:30<04:42, 13.54it/s] Update steps: 62%|██████████████▊ | 6183/10000 [11:30<04:38, 13.70it/s] Update steps: 62%|██████████████▊ | 6185/10000 [11:30<04:39, 13.64it/s] Update steps: 62%|██████████████▊ | 6187/10000 [11:30<04:38, 13.68it/s] Update steps: 62%|██████████████▊ | 6189/10000 [11:30<04:38, 13.70it/s] Update steps: 62%|██████████████▊ | 6191/10000 [11:31<04:41, 13.54it/s] Update steps: 62%|██████████████▊ | 6193/10000 [11:31<04:38, 13.68it/s] Update steps: 62%|██████████████▊ | 6195/10000 [11:31<04:36, 13.77it/s] Update steps: 62%|██████████████▊ | 6197/10000 [11:31<04:35, 13.79it/s] Update steps: 62%|██████████████▉ | 6199/10000 [11:31<04:39, 13.58it/s] Update steps: 62%|██████████████▉ | 6201/10000 [11:31<04:37, 13.69it/s] Update steps: 62%|██████████████▉ | 6203/10000 [11:31<04:36, 13.73it/s] Update steps: 62%|██████████████▉ | 6205/10000 [11:32<04:37, 13.67it/s] Update steps: 62%|██████████████▉ | 6207/10000 [11:32<04:39, 13.58it/s] Update steps: 62%|██████████████▉ | 6209/10000 [11:32<04:36, 13.70it/s] Update steps: 62%|██████████████▉ | 6211/10000 [11:32<04:38, 13.59it/s] Update steps: 62%|██████████████▉ | 6213/10000 [11:32<04:36, 13.70it/s] Update steps: 62%|██████████████▉ | 6215/10000 [11:32<04:35, 13.72it/s] Update steps: 62%|██████████████▉ | 6217/10000 [11:32<04:34, 13.76it/s] Update steps: 62%|██████████████▉ | 6219/10000 [11:33<04:32, 13.85it/s] Update steps: 62%|██████████████▉ | 6221/10000 [11:33<04:32, 13.88it/s] Update steps: 62%|██████████████▉ | 6223/10000 [11:33<04:32, 13.87it/s] Update steps: 62%|██████████████▉ | 6225/10000 [11:33<04:35, 13.71it/s] Update steps: 62%|██████████████▉ | 6227/10000 [11:33<04:35, 13.70it/s] Update steps: 62%|██████████████▉ | 6229/10000 [11:33<04:33, 13.81it/s] Update steps: 62%|██████████████▉ | 6231/10000 [11:33<04:34, 13.71it/s] Update steps: 62%|██████████████▉ | 6233/10000 [11:34<04:34, 13.72it/s] Update steps: 62%|██████████████▉ | 6235/10000 [11:34<04:34, 13.73it/s] Update steps: 62%|██████████████▉ | 6237/10000 [11:34<04:35, 13.64it/s] Update steps: 62%|██████████████▉ | 6239/10000 [11:34<04:34, 13.70it/s] Update steps: 62%|██████████████▉ | 6241/10000 [11:34<04:31, 13.83it/s] Update steps: 62%|██████████████▉ | 6243/10000 [11:34<04:34, 13.68it/s] Update steps: 62%|██████████████▉ | 6245/10000 [11:34<04:33, 13.74it/s] Update steps: 62%|██████████████▉ | 6247/10000 [11:35<04:37, 13.55it/s] Update steps: 62%|██████████████▉ | 6249/10000 [11:35<04:33, 13.70it/s] Update steps: 63%|███████████████ | 6251/10000 [11:35<04:33, 13.71it/s] Update steps: 63%|███████████████ | 6253/10000 [11:35<04:32, 13.74it/s] Update steps: 63%|███████████████ | 6255/10000 [11:35<04:34, 13.63it/s] Update steps: 63%|███████████████ | 6257/10000 [11:35<04:32, 13.72it/s] Update steps: 63%|███████████████ | 6259/10000 [11:35<04:31, 13.78it/s] Update steps: 63%|███████████████ | 6261/10000 [11:36<04:33, 13.66it/s] Update steps: 63%|███████████████ | 6263/10000 [11:36<04:32, 13.71it/s] Update steps: 63%|███████████████ | 6265/10000 [11:36<04:31, 13.76it/s] Update steps: 63%|███████████████ | 6267/10000 [11:36<04:31, 13.74it/s] Update steps: 63%|███████████████ | 6269/10000 [11:36<04:31, 13.74it/s] Update steps: 63%|███████████████ | 6271/10000 [11:36<04:33, 13.65it/s] Update steps: 63%|███████████████ | 6273/10000 [11:37<04:33, 13.62it/s] Update steps: 63%|███████████████ | 6275/10000 [11:37<04:34, 13.57it/s] Update steps: 63%|███████████████ | 6277/10000 [11:37<04:34, 13.57it/s] Update steps: 63%|███████████████ | 6279/10000 [11:37<04:34, 13.58it/s] Update steps: 63%|███████████████ | 6281/10000 [11:37<04:32, 13.67it/s] Update steps: 63%|███████████████ | 6283/10000 [11:37<04:30, 13.76it/s] Update steps: 63%|███████████████ | 6285/10000 [11:37<04:30, 13.75it/s] Update steps: 63%|███████████████ | 6287/10000 [11:38<04:34, 13.54it/s] Update steps: 63%|███████████████ | 6289/10000 [11:38<04:31, 13.67it/s] Update steps: 63%|███████████████ | 6291/10000 [11:38<04:30, 13.73it/s] Update steps: 63%|███████████████ | 6293/10000 [11:38<04:29, 13.77it/s] Update steps: 63%|███████████████ | 6295/10000 [11:38<04:28, 13.78it/s] Update steps: 63%|███████████████ | 6297/10000 [11:38<04:27, 13.83it/s] Update steps: 63%|███████████████ | 6299/10000 [11:38<04:27, 13.85it/s] Update steps: 63%|███████████████ | 6301/10000 [11:39<04:26, 13.91it/s] Update steps: 63%|███████████████▏ | 6303/10000 [11:39<04:25, 13.93it/s] Update steps: 63%|███████████████▏ | 6305/10000 [11:39<04:25, 13.92it/s] Update steps: 63%|███████████████▏ | 6307/10000 [11:39<04:24, 13.96it/s] Update steps: 63%|███████████████▏ | 6309/10000 [11:39<04:27, 13.82it/s] Update steps: 63%|███████████████▏ | 6311/10000 [11:39<04:36, 13.35it/s] Update steps: 63%|███████████████▏ | 6313/10000 [11:39<04:32, 13.51it/s] Update steps: 63%|███████████████▏ | 6315/10000 [11:40<04:31, 13.58it/s] Update steps: 63%|███████████████▏ | 6317/10000 [11:40<04:31, 13.56it/s] Update steps: 63%|███████████████▏ | 6319/10000 [11:40<04:29, 13.67it/s] Update steps: 63%|███████████████▏ | 6321/10000 [11:40<04:33, 13.46it/s] Update steps: 63%|███████████████▏ | 6323/10000 [11:40<04:31, 13.54it/s] Update steps: 63%|███████████████▏ | 6325/10000 [11:40<04:30, 13.56it/s] Update steps: 63%|███████████████▏ | 6327/10000 [11:40<04:29, 13.65it/s] Update steps: 63%|███████████████▏ | 6329/10000 [11:41<04:27, 13.72it/s] Update steps: 63%|███████████████▏ | 6331/10000 [11:41<04:25, 13.81it/s] Update steps: 63%|███████████████▏ | 6333/10000 [11:41<04:24, 13.86it/s] Update steps: 63%|███████████████▏ | 6335/10000 [11:41<04:23, 13.90it/s] Update steps: 63%|███████████████▏ | 6337/10000 [11:41<04:22, 13.95it/s] Update steps: 63%|███████████████▏ | 6339/10000 [11:41<04:22, 13.92it/s] Update steps: 63%|███████████████▏ | 6341/10000 [11:41<04:23, 13.89it/s] Update steps: 63%|███████████████▏ | 6343/10000 [11:42<04:29, 13.56it/s] Update steps: 63%|███████████████▏ | 6345/10000 [11:42<04:27, 13.67it/s] Update steps: 63%|███████████████▏ | 6347/10000 [11:42<04:25, 13.77it/s] Update steps: 63%|███████████████▏ | 6349/10000 [11:42<04:24, 13.80it/s] Update steps: 64%|███████████████▏ | 6351/10000 [11:42<04:27, 13.66it/s] Update steps: 64%|███████████████▏ | 6353/10000 [11:42<04:24, 13.77it/s] Update steps: 64%|███████████████▎ | 6355/10000 [11:42<04:24, 13.76it/s] Update steps: 64%|███████████████▎ | 6357/10000 [11:43<04:26, 13.68it/s] Update steps: 64%|███████████████▎ | 6359/10000 [11:43<04:25, 13.69it/s] Update steps: 64%|███████████████▎ | 6361/10000 [11:43<04:32, 13.36it/s] Update steps: 64%|███████████████▎ | 6363/10000 [11:43<04:29, 13.47it/s] Update steps: 64%|███████████████▎ | 6365/10000 [11:43<04:30, 13.43it/s] Update steps: 64%|███████████████▎ | 6367/10000 [11:43<04:28, 13.53it/s] Update steps: 64%|███████████████▎ | 6369/10000 [11:44<04:25, 13.70it/s] Update steps: 64%|███████████████▎ | 6371/10000 [11:44<04:24, 13.73it/s] Update steps: 64%|███████████████▎ | 6373/10000 [11:44<04:23, 13.78it/s] Update steps: 64%|███████████████▎ | 6375/10000 [11:44<04:24, 13.71it/s] Update steps: 64%|███████████████▎ | 6377/10000 [11:44<04:23, 13.76it/s] Update steps: 64%|███████████████▎ | 6379/10000 [11:44<04:21, 13.87it/s] Update steps: 64%|███████████████▎ | 6381/10000 [11:44<04:20, 13.88it/s] Update steps: 64%|███████████████▎ | 6383/10000 [11:45<04:21, 13.86it/s] Update steps: 64%|███████████████▎ | 6385/10000 [11:45<04:21, 13.84it/s] Update steps: 64%|███████████████▎ | 6387/10000 [11:45<04:29, 13.40it/s] Update steps: 64%|███████████████▎ | 6389/10000 [11:45<04:28, 13.43it/s] Update steps: 64%|███████████████▎ | 6391/10000 [11:45<04:26, 13.54it/s] Update steps: 64%|███████████████▎ | 6393/10000 [11:45<04:23, 13.67it/s] Update steps: 64%|███████████████▎ | 6395/10000 [11:45<04:22, 13.75it/s] Update steps: 64%|███████████████▎ | 6397/10000 [11:46<04:20, 13.83it/s] Update steps: 64%|███████████████▎ | 6399/10000 [11:46<04:19, 13.90it/s] Update steps: 64%|███████████████▎ | 6401/10000 [11:46<04:21, 13.76it/s] Update steps: 64%|███████████████▎ | 6403/10000 [11:46<04:19, 13.85it/s] Update steps: 64%|███████████████▎ | 6405/10000 [11:46<04:18, 13.90it/s] Update steps: 64%|███████████████▍ | 6407/10000 [11:46<04:18, 13.92it/s] Update steps: 64%|███████████████▍ | 6409/10000 [11:46<04:16, 13.99it/s] Update steps: 64%|███████████████▍ | 6411/10000 [11:47<04:17, 13.93it/s] Update steps: 64%|███████████████▍ | 6413/10000 [11:47<04:17, 13.95it/s] Update steps: 64%|███████████████▍ | 6415/10000 [11:47<04:22, 13.67it/s] Update steps: 64%|███████████████▍ | 6417/10000 [11:47<04:20, 13.77it/s] Update steps: 64%|███████████████▍ | 6419/10000 [11:47<04:19, 13.80it/s] Update steps: 64%|███████████████▍ | 6421/10000 [11:47<04:20, 13.73it/s] Update steps: 64%|███████████████▍ | 6423/10000 [11:47<04:31, 13.18it/s] Update steps: 64%|███████████████▍ | 6425/10000 [11:48<04:28, 13.31it/s] Update steps: 64%|███████████████▍ | 6427/10000 [11:48<04:27, 13.34it/s] Update steps: 64%|███████████████▍ | 6429/10000 [11:48<04:23, 13.57it/s] Update steps: 64%|███████████████▍ | 6431/10000 [11:48<04:21, 13.67it/s] Update steps: 64%|███████████████▍ | 6433/10000 [11:48<04:20, 13.69it/s] Update steps: 64%|███████████████▍ | 6435/10000 [11:48<04:20, 13.69it/s] Update steps: 64%|███████████████▍ | 6437/10000 [11:48<04:19, 13.75it/s] Update steps: 64%|███████████████▍ | 6439/10000 [11:49<04:18, 13.80it/s] Update steps: 64%|███████████████▍ | 6441/10000 [11:49<04:18, 13.74it/s] Update steps: 64%|███████████████▍ | 6443/10000 [11:49<04:20, 13.67it/s] Update steps: 64%|███████████████▍ | 6445/10000 [11:49<04:17, 13.79it/s] Update steps: 64%|███████████████▍ | 6447/10000 [11:49<04:18, 13.75it/s] Update steps: 64%|███████████████▍ | 6449/10000 [11:49<04:17, 13.77it/s] Update steps: 65%|███████████████▍ | 6451/10000 [11:50<04:16, 13.83it/s] Update steps: 65%|███████████████▍ | 6453/10000 [11:50<04:21, 13.58it/s] Update steps: 65%|███████████████▍ | 6455/10000 [11:50<04:21, 13.58it/s] Update steps: 65%|███████████████▍ | 6457/10000 [11:50<04:21, 13.54it/s] Update steps: 65%|███████████████▌ | 6459/10000 [11:50<04:21, 13.57it/s] Update steps: 65%|███████████████▌ | 6461/10000 [11:50<04:22, 13.50it/s] Update steps: 65%|███████████████▌ | 6463/10000 [11:50<04:22, 13.50it/s] Update steps: 65%|███████████████▌ | 6465/10000 [11:51<04:18, 13.67it/s] Update steps: 65%|███████████████▌ | 6467/10000 [11:51<04:18, 13.68it/s] Update steps: 65%|███████████████▌ | 6469/10000 [11:51<04:18, 13.65it/s] Update steps: 65%|███████████████▌ | 6471/10000 [11:51<04:18, 13.67it/s] Update steps: 65%|███████████████▌ | 6473/10000 [11:51<04:16, 13.74it/s] Update steps: 65%|███████████████▌ | 6475/10000 [11:51<04:16, 13.72it/s] Update steps: 65%|███████████████▌ | 6477/10000 [11:51<04:18, 13.61it/s] Update steps: 65%|███████████████▌ | 6479/10000 [11:52<04:20, 13.50it/s] Update steps: 65%|███████████████▌ | 6481/10000 [11:52<04:20, 13.51it/s] Update steps: 65%|███████████████▌ | 6483/10000 [11:52<04:18, 13.61it/s] Update steps: 65%|███████████████▌ | 6485/10000 [11:52<04:16, 13.68it/s] Update steps: 65%|███████████████▌ | 6487/10000 [11:52<04:14, 13.79it/s] Update steps: 65%|███████████████▌ | 6489/10000 [11:52<04:13, 13.83it/s] Update steps: 65%|███████████████▌ | 6491/10000 [11:52<04:14, 13.80it/s] Update steps: 65%|███████████████▌ | 6493/10000 [11:53<04:13, 13.82it/s] Update steps: 65%|███████████████▌ | 6495/10000 [11:53<04:12, 13.86it/s] Update steps: 65%|███████████████▌ | 6497/10000 [11:53<04:12, 13.85it/s] Update steps: 65%|███████████████▌ | 6499/10000 [11:53<04:17, 13.58it/s] Update steps: 65%|███████████████▌ | 6501/10000 [11:53<04:15, 13.70it/s] Update steps: 65%|███████████████▌ | 6503/10000 [11:53<04:14, 13.74it/s] Update steps: 65%|███████████████▌ | 6505/10000 [11:53<04:14, 13.74it/s] Update steps: 65%|███████████████▌ | 6507/10000 [11:54<04:13, 13.76it/s] Update steps: 65%|███████████████▌ | 6509/10000 [11:54<04:14, 13.74it/s] Update steps: 65%|███████████████▋ | 6511/10000 [11:54<04:14, 13.70it/s] Update steps: 65%|███████████████▋ | 6513/10000 [11:54<04:13, 13.76it/s] Update steps: 65%|███████████████▋ | 6515/10000 [11:54<04:13, 13.75it/s] Update steps: 65%|███████████████▋ | 6517/10000 [11:54<04:13, 13.75it/s] Update steps: 65%|███████████████▋ | 6519/10000 [11:54<04:13, 13.73it/s] Update steps: 65%|███████████████▋ | 6521/10000 [11:55<04:12, 13.77it/s] Update steps: 65%|███████████████▋ | 6523/10000 [11:55<04:11, 13.84it/s] Update steps: 65%|███████████████▋ | 6525/10000 [11:55<04:11, 13.80it/s] Update steps: 65%|███████████████▋ | 6527/10000 [11:55<04:12, 13.76it/s] Update steps: 65%|███████████████▋ | 6529/10000 [11:55<04:13, 13.67it/s] Update steps: 65%|███████████████▋ | 6531/10000 [11:55<04:13, 13.70it/s] Update steps: 65%|███████████████▋ | 6533/10000 [11:55<04:12, 13.72it/s] Update steps: 65%|███████████████▋ | 6535/10000 [11:56<04:13, 13.67it/s] Update steps: 65%|███████████████▋ | 6537/10000 [11:56<04:12, 13.74it/s] Update steps: 65%|███████████████▋ | 6539/10000 [11:56<04:11, 13.78it/s] Update steps: 65%|███████████████▋ | 6541/10000 [11:56<04:11, 13.74it/s] Update steps: 65%|███████████████▋ | 6543/10000 [11:56<04:12, 13.69it/s] Update steps: 65%|███████████████▋ | 6545/10000 [11:56<04:10, 13.79it/s] Update steps: 65%|███████████████▋ | 6547/10000 [11:57<04:11, 13.73it/s] Update steps: 65%|███████████████▋ | 6549/10000 [11:57<04:10, 13.80it/s] Update steps: 66%|███████████████▋ | 6551/10000 [11:57<04:10, 13.78it/s] Update steps: 66%|███████████████▋ | 6553/10000 [11:57<04:08, 13.87it/s] Update steps: 66%|███████████████▋ | 6555/10000 [11:57<04:09, 13.82it/s] Update steps: 66%|███████████████▋ | 6557/10000 [11:57<04:09, 13.81it/s] Update steps: 66%|███████████████▋ | 6559/10000 [11:57<04:12, 13.65it/s] Update steps: 66%|███████████████▋ | 6561/10000 [11:58<04:12, 13.61it/s] Update steps: 66%|███████████████▊ | 6563/10000 [11:58<04:10, 13.72it/s] Update steps: 66%|███████████████▊ | 6565/10000 [11:58<04:10, 13.72it/s] Update steps: 66%|███████████████▊ | 6567/10000 [11:58<04:08, 13.79it/s] Update steps: 66%|███████████████▊ | 6569/10000 [11:58<04:07, 13.86it/s] Update steps: 66%|███████████████▊ | 6571/10000 [11:58<04:09, 13.76it/s] Update steps: 66%|███████████████▊ | 6573/10000 [11:58<04:08, 13.79it/s] Update steps: 66%|███████████████▊ | 6575/10000 [11:59<04:10, 13.69it/s] Update steps: 66%|███████████████▊ | 6577/10000 [11:59<04:10, 13.65it/s] Update steps: 66%|███████████████▊ | 6579/10000 [11:59<04:11, 13.61it/s] Update steps: 66%|███████████████▊ | 6581/10000 [11:59<04:09, 13.68it/s] Update steps: 66%|███████████████▊ | 6583/10000 [11:59<04:07, 13.78it/s] Update steps: 66%|███████████████▊ | 6585/10000 [11:59<04:09, 13.68it/s] Update steps: 66%|███████████████▊ | 6587/10000 [11:59<04:09, 13.67it/s] Update steps: 66%|███████████████▊ | 6589/10000 [12:00<04:08, 13.75it/s] Update steps: 66%|███████████████▊ | 6591/10000 [12:00<04:07, 13.76it/s] Update steps: 66%|███████████████▊ | 6593/10000 [12:00<04:07, 13.77it/s] Update steps: 66%|███████████████▊ | 6595/10000 [12:00<04:11, 13.55it/s] Update steps: 66%|███████████████▊ | 6597/10000 [12:00<04:10, 13.61it/s] Update steps: 66%|███████████████▊ | 6599/10000 [12:00<04:08, 13.69it/s] Update steps: 66%|███████████████▊ | 6601/10000 [12:00<04:09, 13.62it/s] Update steps: 66%|███████████████▊ | 6603/10000 [12:01<04:09, 13.61it/s] Update steps: 66%|███████████████▊ | 6605/10000 [12:01<04:07, 13.74it/s] Update steps: 66%|███████████████▊ | 6607/10000 [12:01<04:09, 13.63it/s] Update steps: 66%|███████████████▊ | 6609/10000 [12:01<04:11, 13.50it/s] Update steps: 66%|███████████████▊ | 6611/10000 [12:01<04:10, 13.52it/s] Update steps: 66%|███████████████▊ | 6613/10000 [12:01<04:09, 13.55it/s] Update steps: 66%|███████████████▉ | 6615/10000 [12:01<04:08, 13.61it/s] Update steps: 66%|███████████████▉ | 6617/10000 [12:02<04:06, 13.73it/s] Update steps: 66%|███████████████▉ | 6619/10000 [12:02<04:04, 13.82it/s] Update steps: 66%|███████████████▉ | 6621/10000 [12:02<04:03, 13.87it/s] Update steps: 66%|███████████████▉ | 6623/10000 [12:02<04:02, 13.94it/s] Update steps: 66%|███████████████▉ | 6625/10000 [12:02<04:01, 14.00it/s] Update steps: 66%|███████████████▉ | 6627/10000 [12:02<04:01, 13.98it/s] Update steps: 66%|███████████████▉ | 6629/10000 [12:02<04:02, 13.88it/s] Update steps: 66%|███████████████▉ | 6631/10000 [12:03<04:02, 13.87it/s] Update steps: 66%|███████████████▉ | 6633/10000 [12:03<04:05, 13.71it/s] Update steps: 66%|███████████████▉ | 6635/10000 [12:03<04:03, 13.80it/s] Update steps: 66%|███████████████▉ | 6637/10000 [12:03<04:03, 13.82it/s] Update steps: 66%|███████████████▉ | 6639/10000 [12:03<04:11, 13.36it/s] Update steps: 66%|███████████████▉ | 6641/10000 [12:03<04:11, 13.38it/s] Update steps: 66%|███████████████▉ | 6643/10000 [12:04<04:06, 13.59it/s] Update steps: 66%|███████████████▉ | 6645/10000 [12:04<04:04, 13.73it/s] Update steps: 66%|███████████████▉ | 6647/10000 [12:04<04:03, 13.79it/s] Update steps: 66%|███████████████▉ | 6649/10000 [12:04<04:01, 13.85it/s] Update steps: 67%|███████████████▉ | 6651/10000 [12:04<04:01, 13.87it/s] Update steps: 67%|███████████████▉ | 6653/10000 [12:04<04:00, 13.92it/s] Update steps: 67%|███████████████▉ | 6655/10000 [12:04<04:01, 13.83it/s] Update steps: 67%|███████████████▉ | 6657/10000 [12:05<04:00, 13.89it/s] Update steps: 67%|███████████████▉ | 6659/10000 [12:05<04:01, 13.82it/s] Update steps: 67%|███████████████▉ | 6661/10000 [12:05<04:02, 13.79it/s] Update steps: 67%|███████████████▉ | 6663/10000 [12:05<04:00, 13.87it/s] Update steps: 67%|███████████████▉ | 6665/10000 [12:05<04:00, 13.85it/s] Update steps: 67%|████████████████ | 6667/10000 [12:05<04:02, 13.76it/s] Update steps: 67%|████████████████ | 6669/10000 [12:05<04:02, 13.76it/s] Update steps: 67%|████████████████ | 6671/10000 [12:06<04:01, 13.76it/s] Update steps: 67%|████████████████ | 6673/10000 [12:06<04:01, 13.76it/s] Update steps: 67%|████████████████ | 6675/10000 [12:06<04:01, 13.78it/s] Update steps: 67%|████████████████ | 6677/10000 [12:06<04:00, 13.82it/s] Update steps: 67%|████████████████ | 6679/10000 [12:06<04:00, 13.80it/s] Update steps: 67%|████████████████ | 6681/10000 [12:06<04:00, 13.78it/s] Update steps: 67%|████████████████ | 6683/10000 [12:06<03:59, 13.83it/s] Update steps: 67%|████████████████ | 6685/10000 [12:07<04:00, 13.76it/s] Update steps: 67%|████████████████ | 6687/10000 [12:07<03:59, 13.83it/s] Update steps: 67%|████████████████ | 6689/10000 [12:07<03:58, 13.89it/s] Update steps: 67%|████████████████ | 6691/10000 [12:07<03:58, 13.85it/s] Update steps: 67%|████████████████ | 6693/10000 [12:07<04:00, 13.73it/s] Update steps: 67%|████████████████ | 6695/10000 [12:07<04:02, 13.63it/s] Update steps: 67%|████████████████ | 6697/10000 [12:07<04:00, 13.72it/s] Update steps: 67%|████████████████ | 6699/10000 [12:08<03:59, 13.81it/s] Update steps: 67%|████████████████ | 6701/10000 [12:08<03:57, 13.88it/s] Update steps: 67%|████████████████ | 6703/10000 [12:08<04:02, 13.57it/s] Update steps: 67%|████████████████ | 6705/10000 [12:08<04:05, 13.41it/s] Update steps: 67%|████████████████ | 6707/10000 [12:08<04:03, 13.53it/s] Update steps: 67%|████████████████ | 6709/10000 [12:08<04:02, 13.59it/s] Update steps: 67%|████████████████ | 6711/10000 [12:08<04:00, 13.69it/s] Update steps: 67%|████████████████ | 6713/10000 [12:09<03:59, 13.72it/s] Update steps: 67%|████████████████ | 6715/10000 [12:09<04:00, 13.67it/s] Update steps: 67%|████████████████ | 6717/10000 [12:09<03:59, 13.73it/s] Update steps: 67%|████████████████▏ | 6719/10000 [12:09<03:59, 13.70it/s] Update steps: 67%|████████████████▏ | 6721/10000 [12:09<03:57, 13.79it/s] Update steps: 67%|████████████████▏ | 6723/10000 [12:09<03:56, 13.86it/s] Update steps: 67%|████████████████▏ | 6725/10000 [12:09<03:57, 13.80it/s] Update steps: 67%|████████████████▏ | 6727/10000 [12:10<03:58, 13.72it/s] Update steps: 67%|████████████████▏ | 6729/10000 [12:10<03:57, 13.79it/s] Update steps: 67%|████████████████▏ | 6731/10000 [12:10<03:59, 13.66it/s] Update steps: 67%|████████████████▏ | 6733/10000 [12:10<03:57, 13.76it/s] Update steps: 67%|████████████████▏ | 6735/10000 [12:10<03:55, 13.89it/s] Update steps: 67%|████████████████▏ | 6737/10000 [12:10<03:55, 13.86it/s] Update steps: 67%|████████████████▏ | 6739/10000 [12:10<03:54, 13.91it/s] Update steps: 67%|████████████████▏ | 6741/10000 [12:11<03:55, 13.83it/s] Update steps: 67%|████████████████▏ | 6743/10000 [12:11<03:54, 13.87it/s] Update steps: 67%|████████████████▏ | 6745/10000 [12:11<03:53, 13.92it/s] Update steps: 67%|████████████████▏ | 6747/10000 [12:11<03:56, 13.74it/s] Update steps: 67%|████████████████▏ | 6749/10000 [12:11<03:54, 13.84it/s] Update steps: 68%|████████████████▏ | 6751/10000 [12:11<03:53, 13.90it/s] Update steps: 68%|████████████████▏ | 6753/10000 [12:12<03:56, 13.72it/s] Update steps: 68%|████████████████▏ | 6755/10000 [12:12<03:57, 13.66it/s] Update steps: 68%|████████████████▏ | 6757/10000 [12:12<03:58, 13.61it/s] Update steps: 68%|████████████████▏ | 6759/10000 [12:12<04:01, 13.44it/s] Update steps: 68%|████████████████▏ | 6761/10000 [12:12<03:58, 13.60it/s] Update steps: 68%|████████████████▏ | 6763/10000 [12:12<03:56, 13.68it/s] Update steps: 68%|████████████████▏ | 6765/10000 [12:12<03:55, 13.73it/s] Update steps: 68%|████████████████▏ | 6767/10000 [12:13<03:57, 13.64it/s] Update steps: 68%|████████████████▏ | 6769/10000 [12:13<03:55, 13.70it/s] Update steps: 68%|████████████████▎ | 6771/10000 [12:13<03:53, 13.80it/s] Update steps: 68%|████████████████▎ | 6773/10000 [12:13<03:53, 13.81it/s] Update steps: 68%|████████████████▎ | 6775/10000 [12:13<03:54, 13.78it/s] Update steps: 68%|████████████████▎ | 6777/10000 [12:13<03:53, 13.78it/s] Update steps: 68%|████████████████▎ | 6779/10000 [12:13<03:52, 13.83it/s] Update steps: 68%|████████████████▎ | 6781/10000 [12:14<03:54, 13.72it/s] Update steps: 68%|████████████████▎ | 6783/10000 [12:14<03:53, 13.75it/s] Update steps: 68%|████████████████▎ | 6785/10000 [12:14<03:52, 13.83it/s] Update steps: 68%|████████████████▎ | 6787/10000 [12:14<03:52, 13.79it/s] Update steps: 68%|████████████████▎ | 6789/10000 [12:14<03:51, 13.87it/s] Update steps: 68%|████████████████▎ | 6791/10000 [12:14<03:50, 13.91it/s] Update steps: 68%|████████████████▎ | 6793/10000 [12:14<03:49, 13.94it/s] Update steps: 68%|████████████████▎ | 6795/10000 [12:15<03:49, 13.97it/s] Update steps: 68%|████████████████▎ | 6797/10000 [12:15<03:48, 13.99it/s] Update steps: 68%|████████████████▎ | 6799/10000 [12:15<03:49, 13.97it/s] Update steps: 68%|████████████████▎ | 6801/10000 [12:15<03:50, 13.88it/s] Update steps: 68%|████████████████▎ | 6803/10000 [12:15<03:52, 13.78it/s] Update steps: 68%|████████████████▎ | 6805/10000 [12:15<03:51, 13.83it/s] Update steps: 68%|████████████████▎ | 6807/10000 [12:15<03:49, 13.92it/s] Update steps: 68%|████████████████▎ | 6809/10000 [12:16<03:49, 13.91it/s] Update steps: 68%|████████████████▎ | 6811/10000 [12:16<03:50, 13.86it/s] Update steps: 68%|████████████████▎ | 6813/10000 [12:16<03:50, 13.82it/s] Update steps: 68%|████████████████▎ | 6815/10000 [12:16<03:50, 13.81it/s] Update steps: 68%|████████████████▎ | 6817/10000 [12:16<03:50, 13.79it/s] Update steps: 68%|████████████████▎ | 6819/10000 [12:16<03:51, 13.77it/s] Update steps: 68%|████████████████▎ | 6821/10000 [12:16<03:50, 13.79it/s] Update steps: 68%|████████████████▍ | 6823/10000 [12:17<03:50, 13.78it/s] Update steps: 68%|████████████████▍ | 6825/10000 [12:17<03:49, 13.82it/s] Update steps: 68%|████████████████▍ | 6827/10000 [12:17<03:49, 13.82it/s] Update steps: 68%|████████████████▍ | 6829/10000 [12:17<03:49, 13.83it/s] Update steps: 68%|████████████████▍ | 6831/10000 [12:17<03:49, 13.79it/s] Update steps: 68%|████████████████▍ | 6833/10000 [12:17<03:50, 13.74it/s] Update steps: 68%|████████████████▍ | 6835/10000 [12:17<03:50, 13.72it/s] Update steps: 68%|████████████████▍ | 6837/10000 [12:18<03:48, 13.86it/s] Update steps: 68%|████████████████▍ | 6839/10000 [12:18<03:47, 13.90it/s] Update steps: 68%|████████████████▍ | 6841/10000 [12:18<03:46, 13.95it/s] Update steps: 68%|████████████████▍ | 6843/10000 [12:18<03:47, 13.89it/s] Update steps: 68%|████████████████▍ | 6845/10000 [12:18<03:50, 13.68it/s] Update steps: 68%|████████████████▍ | 6847/10000 [12:18<03:50, 13.68it/s] Update steps: 68%|████████████████▍ | 6849/10000 [12:18<03:50, 13.68it/s] Update steps: 69%|████████████████▍ | 6851/10000 [12:19<03:49, 13.75it/s] Update steps: 69%|████████████████▍ | 6853/10000 [12:19<03:47, 13.84it/s] Update steps: 69%|████████████████▍ | 6855/10000 [12:19<03:48, 13.78it/s] Update steps: 69%|████████████████▍ | 6857/10000 [12:19<03:46, 13.85it/s] Update steps: 69%|████████████████▍ | 6859/10000 [12:19<03:47, 13.81it/s] Update steps: 69%|████████████████▍ | 6861/10000 [12:19<03:48, 13.76it/s] Update steps: 69%|████████████████▍ | 6863/10000 [12:19<03:46, 13.84it/s] Update steps: 69%|████████████████▍ | 6865/10000 [12:20<03:46, 13.86it/s] Update steps: 69%|████████████████▍ | 6867/10000 [12:20<03:45, 13.91it/s] Update steps: 69%|████████████████▍ | 6869/10000 [12:20<03:45, 13.91it/s] Update steps: 69%|████████████████▍ | 6871/10000 [12:20<03:46, 13.84it/s] Update steps: 69%|████████████████▍ | 6873/10000 [12:20<03:46, 13.83it/s] Update steps: 69%|████████████████▌ | 6875/10000 [12:20<03:48, 13.67it/s] Update steps: 69%|████████████████▌ | 6877/10000 [12:20<03:46, 13.77it/s] Update steps: 69%|████████████████▌ | 6879/10000 [12:21<03:45, 13.81it/s] Update steps: 69%|████████████████▌ | 6881/10000 [12:21<03:51, 13.50it/s] Update steps: 69%|████████████████▌ | 6883/10000 [12:21<03:51, 13.44it/s] Update steps: 69%|████████████████▌ | 6885/10000 [12:21<03:51, 13.45it/s] Update steps: 69%|████████████████▌ | 6887/10000 [12:21<03:50, 13.53it/s] Update steps: 69%|████████████████▌ | 6889/10000 [12:21<03:53, 13.34it/s] Update steps: 69%|████████████████▌ | 6891/10000 [12:22<03:54, 13.24it/s] Update steps: 69%|████████████████▌ | 6893/10000 [12:22<03:52, 13.35it/s] Update steps: 69%|████████████████▌ | 6895/10000 [12:22<03:50, 13.50it/s] Update steps: 69%|████████████████▌ | 6897/10000 [12:22<03:47, 13.63it/s] Update steps: 69%|████████████████▌ | 6899/10000 [12:22<03:45, 13.75it/s] Update steps: 69%|████████████████▌ | 6901/10000 [12:22<03:44, 13.83it/s] Update steps: 69%|████████████████▌ | 6903/10000 [12:22<03:42, 13.90it/s] Update steps: 69%|████████████████▌ | 6905/10000 [12:23<03:42, 13.91it/s] Update steps: 69%|████████████████▌ | 6907/10000 [12:23<03:44, 13.81it/s] Update steps: 69%|████████████████▌ | 6909/10000 [12:23<03:42, 13.90it/s] Update steps: 69%|████████████████▌ | 6911/10000 [12:23<03:42, 13.89it/s] Update steps: 69%|████████████████▌ | 6913/10000 [12:23<03:40, 13.99it/s] Update steps: 69%|████████████████▌ | 6915/10000 [12:23<03:40, 13.99it/s] Update steps: 69%|████████████████▌ | 6917/10000 [12:23<03:41, 13.89it/s] Update steps: 69%|████████████████▌ | 6919/10000 [12:24<03:44, 13.70it/s] Update steps: 69%|████████████████▌ | 6921/10000 [12:24<03:44, 13.70it/s] Update steps: 69%|████████████████▌ | 6923/10000 [12:24<03:43, 13.77it/s] Update steps: 69%|████████████████▌ | 6925/10000 [12:24<03:42, 13.84it/s] Update steps: 69%|████████████████▌ | 6927/10000 [12:24<03:42, 13.84it/s] Update steps: 69%|████████████████▋ | 6929/10000 [12:24<03:45, 13.63it/s] Update steps: 69%|████████████████▋ | 6931/10000 [12:24<03:45, 13.59it/s] Update steps: 69%|████████████████▋ | 6933/10000 [12:25<03:44, 13.65it/s] Update steps: 69%|████████████████▋ | 6935/10000 [12:25<03:45, 13.61it/s] Update steps: 69%|████████████████▋ | 6937/10000 [12:25<03:43, 13.72it/s] Update steps: 69%|████████████████▋ | 6939/10000 [12:25<03:43, 13.69it/s] Update steps: 69%|████████████████▋ | 6941/10000 [12:25<03:42, 13.75it/s] Update steps: 69%|████████████████▋ | 6943/10000 [12:25<03:44, 13.60it/s] Update steps: 69%|████████████████▋ | 6945/10000 [12:25<03:42, 13.70it/s] Update steps: 69%|████████████████▋ | 6947/10000 [12:26<03:41, 13.78it/s] Update steps: 69%|████████████████▋ | 6949/10000 [12:26<03:39, 13.87it/s] Update steps: 70%|████████████████▋ | 6951/10000 [12:26<03:38, 13.92it/s] Update steps: 70%|████████████████▋ | 6953/10000 [12:26<03:39, 13.87it/s] Update steps: 70%|████████████████▋ | 6955/10000 [12:26<03:40, 13.80it/s] Update steps: 70%|████████████████▋ | 6957/10000 [12:26<03:43, 13.60it/s] Update steps: 70%|████████████████▋ | 6959/10000 [12:26<03:42, 13.69it/s] Update steps: 70%|████████████████▋ | 6961/10000 [12:27<03:43, 13.59it/s] Update steps: 70%|████████████████▋ | 6963/10000 [12:27<03:42, 13.63it/s] Update steps: 70%|████████████████▋ | 6965/10000 [12:27<03:40, 13.75it/s] Update steps: 70%|████████████████▋ | 6967/10000 [12:27<03:40, 13.75it/s] Update steps: 70%|████████████████▋ | 6969/10000 [12:27<03:40, 13.72it/s] Update steps: 70%|████████████████▋ | 6971/10000 [12:27<03:38, 13.83it/s] Update steps: 70%|████████████████▋ | 6973/10000 [12:27<03:37, 13.89it/s] Update steps: 70%|████████████████▋ | 6975/10000 [12:28<03:36, 13.96it/s] Update steps: 70%|████████████████▋ | 6977/10000 [12:28<03:35, 14.00it/s] Update steps: 70%|████████████████▋ | 6979/10000 [12:28<03:36, 13.94it/s] Update steps: 70%|████████████████▊ | 6981/10000 [12:28<03:35, 14.00it/s] Update steps: 70%|████████████████▊ | 6983/10000 [12:28<03:35, 14.00it/s] Update steps: 70%|████████████████▊ | 6985/10000 [12:28<03:36, 13.96it/s] Update steps: 70%|████████████████▊ | 6987/10000 [12:28<03:36, 13.93it/s] Update steps: 70%|████████████████▊ | 6989/10000 [12:29<03:36, 13.89it/s] Update steps: 70%|████████████████▊ | 6991/10000 [12:29<03:37, 13.83it/s] Update steps: 70%|████████████████▊ | 6993/10000 [12:29<03:37, 13.81it/s] Update steps: 70%|████████████████▊ | 6995/10000 [12:29<03:37, 13.84it/s] Update steps: 70%|████████████████▊ | 6997/10000 [12:29<03:36, 13.85it/s] Update steps: 70%|████████████████▊ | 6999/10000 [12:29<03:35, 13.94it/s]2026-04-17 11:05:50.314 | INFO | __main__:main:490 - Performing evaluation at step 7000 +2026-04-17 11:05:50.318 | INFO | utils.eval:evaluate_model:21 - Loading validation data from 8 local validation files +2026-04-17 11:05:51.072 | INFO | utils.eval:evaluate_model:43 - Loaded validation dataset in 0.76 seconds +2026-04-17 11:05:51.073 | INFO | utils.eval:evaluate_model:60 - Eval set prepared in 0.76 seconds + Update steps: 70%|████████████████▊ | 7000/10000 [12:40<03:35, 13.94it/s]2026-04-17 11:06:26.722 | INFO | __main__:main:505 - Eval loss at step 7000: 3.4552492310288203, Eval perplexity: 31.666179893460317 + Update steps: 70%|███████████████▍ | 7001/10000 [13:06<4:36:33, 5.53s/it] Update steps: 70%|███████████████▍ | 7003/10000 [13:06<3:14:32, 3.89s/it] Update steps: 70%|███████████████▍ | 7005/10000 [13:06<2:17:09, 2.75s/it] Update steps: 70%|███████████████▍ | 7007/10000 [13:06<1:37:01, 1.94s/it] Update steps: 70%|███████████████▍ | 7009/10000 [13:06<1:08:59, 1.38s/it] Update steps: 70%|████████████████▊ | 7011/10000 [13:07<49:20, 1.01it/s] Update steps: 70%|████████████████▊ | 7013/10000 [13:07<35:35, 1.40it/s] Update steps: 70%|████████████████▊ | 7015/10000 [13:07<25:58, 1.92it/s] Update steps: 70%|████████████████▊ | 7017/10000 [13:07<19:13, 2.59it/s] Update steps: 70%|████████████████▊ | 7019/10000 [13:07<14:30, 3.42it/s] Update steps: 70%|████████████████▊ | 7021/10000 [13:07<11:14, 4.42it/s] Update steps: 70%|████████████████▊ | 7023/10000 [13:07<08:56, 5.55it/s] Update steps: 70%|████████████████▊ | 7025/10000 [13:08<07:18, 6.79it/s] Update steps: 70%|████████████████▊ | 7027/10000 [13:08<06:10, 8.03it/s] Update steps: 70%|████████████████▊ | 7029/10000 [13:08<05:26, 9.11it/s] Update steps: 70%|████████████████▊ | 7031/10000 [13:08<04:52, 10.16it/s] Update steps: 70%|████████████████▉ | 7033/10000 [13:08<04:29, 11.00it/s] Update steps: 70%|████████████████▉ | 7035/10000 [13:08<04:14, 11.64it/s] Update steps: 70%|████████████████▉ | 7037/10000 [13:09<04:01, 12.28it/s] Update steps: 70%|████████████████▉ | 7039/10000 [13:09<03:52, 12.71it/s] Update steps: 70%|████████████████▉ | 7041/10000 [13:09<03:50, 12.81it/s] Update steps: 70%|████████████████▉ | 7043/10000 [13:09<03:44, 13.16it/s] Update steps: 70%|████████████████▉ | 7045/10000 [13:09<03:41, 13.35it/s] Update steps: 70%|████████████████▉ | 7047/10000 [13:09<03:40, 13.41it/s] Update steps: 70%|████████████████▉ | 7049/10000 [13:09<03:38, 13.51it/s] Update steps: 71%|████████████████▉ | 7051/10000 [13:10<03:37, 13.59it/s] Update steps: 71%|████████████████▉ | 7053/10000 [13:10<03:37, 13.54it/s] Update steps: 71%|████████████████▉ | 7055/10000 [13:10<03:36, 13.62it/s] Update steps: 71%|████████████████▉ | 7057/10000 [13:10<03:33, 13.76it/s] Update steps: 71%|████████████████▉ | 7059/10000 [13:10<03:37, 13.50it/s] Update steps: 71%|████████████████▉ | 7061/10000 [13:10<03:35, 13.62it/s] Update steps: 71%|████████████████▉ | 7063/10000 [13:10<03:33, 13.74it/s] Update steps: 71%|████████████████▉ | 7065/10000 [13:11<03:32, 13.82it/s] Update steps: 71%|████████████████▉ | 7067/10000 [13:11<03:32, 13.79it/s] Update steps: 71%|████████████████▉ | 7069/10000 [13:11<03:34, 13.68it/s] Update steps: 71%|████████████████▉ | 7071/10000 [13:11<03:33, 13.74it/s] Update steps: 71%|████████████████▉ | 7073/10000 [13:11<03:32, 13.77it/s] Update steps: 71%|████████████████▉ | 7075/10000 [13:11<03:31, 13.83it/s] Update steps: 71%|████████████████▉ | 7077/10000 [13:11<03:29, 13.94it/s] Update steps: 71%|████████████████▉ | 7079/10000 [13:12<03:29, 13.92it/s] Update steps: 71%|████████████████▉ | 7081/10000 [13:12<03:29, 13.95it/s] Update steps: 71%|████████████████▉ | 7083/10000 [13:12<03:28, 13.99it/s] Update steps: 71%|█████████████████ | 7085/10000 [13:12<03:28, 14.00it/s] Update steps: 71%|█████████████████ | 7087/10000 [13:12<03:33, 13.63it/s] Update steps: 71%|█████████████████ | 7089/10000 [13:12<03:32, 13.69it/s] Update steps: 71%|█████████████████ | 7091/10000 [13:12<03:31, 13.75it/s] Update steps: 71%|█████████████████ | 7093/10000 [13:13<03:32, 13.69it/s] Update steps: 71%|█████████████████ | 7095/10000 [13:13<03:31, 13.74it/s] Update steps: 71%|█████████████████ | 7097/10000 [13:13<03:29, 13.84it/s] Update steps: 71%|█████████████████ | 7099/10000 [13:13<03:28, 13.91it/s] Update steps: 71%|█████████████████ | 7101/10000 [13:13<03:27, 13.95it/s] Update steps: 71%|█████████████████ | 7103/10000 [13:13<03:28, 13.92it/s] Update steps: 71%|█████████████████ | 7105/10000 [13:13<03:29, 13.84it/s] Update steps: 71%|█████████████████ | 7107/10000 [13:14<03:29, 13.83it/s] Update steps: 71%|█████████████████ | 7109/10000 [13:14<03:29, 13.77it/s] Update steps: 71%|█████████████████ | 7111/10000 [13:14<03:28, 13.88it/s] Update steps: 71%|█████████████████ | 7113/10000 [13:14<03:26, 13.97it/s] Update steps: 71%|█████████████████ | 7115/10000 [13:14<03:26, 13.99it/s] Update steps: 71%|█████████████████ | 7117/10000 [13:14<03:26, 13.98it/s] Update steps: 71%|█████████████████ | 7119/10000 [13:14<03:28, 13.80it/s] Update steps: 71%|█████████████████ | 7121/10000 [13:15<03:31, 13.61it/s] Update steps: 71%|█████████████████ | 7123/10000 [13:15<03:30, 13.65it/s] Update steps: 71%|█████████████████ | 7125/10000 [13:15<03:29, 13.72it/s] Update steps: 71%|█████████████████ | 7127/10000 [13:15<03:28, 13.75it/s] Update steps: 71%|█████████████████ | 7129/10000 [13:15<03:27, 13.85it/s] Update steps: 71%|█████████████████ | 7131/10000 [13:15<03:26, 13.92it/s] Update steps: 71%|█████████████████ | 7133/10000 [13:15<03:26, 13.91it/s] Update steps: 71%|█████████████████ | 7135/10000 [13:16<03:26, 13.89it/s] Update steps: 71%|█████████████████▏ | 7137/10000 [13:16<03:25, 13.94it/s] Update steps: 71%|█████████████████▏ | 7139/10000 [13:16<03:26, 13.87it/s] Update steps: 71%|█████████████████▏ | 7141/10000 [13:16<03:27, 13.80it/s] Update steps: 71%|█████████████████▏ | 7143/10000 [13:16<03:27, 13.79it/s] Update steps: 71%|█████████████████▏ | 7145/10000 [13:16<03:27, 13.78it/s] Update steps: 71%|█████████████████▏ | 7147/10000 [13:16<03:27, 13.76it/s] Update steps: 71%|█████████████████▏ | 7149/10000 [13:17<03:28, 13.67it/s] Update steps: 72%|█████████████████▏ | 7151/10000 [13:17<03:26, 13.78it/s] Update steps: 72%|█████████████████▏ | 7153/10000 [13:17<03:30, 13.55it/s] Update steps: 72%|█████████████████▏ | 7155/10000 [13:17<03:27, 13.68it/s] Update steps: 72%|█████████████████▏ | 7157/10000 [13:17<03:26, 13.74it/s] Update steps: 72%|█████████████████▏ | 7159/10000 [13:17<03:25, 13.79it/s] Update steps: 72%|█████████████████▏ | 7161/10000 [13:18<03:25, 13.81it/s] Update steps: 72%|█████████████████▏ | 7163/10000 [13:18<03:24, 13.89it/s] Update steps: 72%|█████████████████▏ | 7165/10000 [13:18<03:23, 13.94it/s] Update steps: 72%|█████████████████▏ | 7167/10000 [13:18<03:24, 13.88it/s] Update steps: 72%|█████████████████▏ | 7169/10000 [13:18<03:25, 13.80it/s] Update steps: 72%|█████████████████▏ | 7171/10000 [13:18<03:26, 13.70it/s] Update steps: 72%|█████████████████▏ | 7173/10000 [13:18<03:25, 13.73it/s] Update steps: 72%|█████████████████▏ | 7175/10000 [13:19<03:27, 13.60it/s] Update steps: 72%|█████████████████▏ | 7177/10000 [13:19<03:28, 13.52it/s] Update steps: 72%|█████████████████▏ | 7179/10000 [13:19<03:27, 13.63it/s] Update steps: 72%|█████████████████▏ | 7181/10000 [13:19<03:25, 13.75it/s] Update steps: 72%|█████████████████▏ | 7183/10000 [13:19<03:23, 13.85it/s] Update steps: 72%|█████████████████▏ | 7185/10000 [13:19<03:23, 13.83it/s] Update steps: 72%|█████████████████▏ | 7187/10000 [13:19<03:22, 13.86it/s] Update steps: 72%|█████████████████▎ | 7189/10000 [13:20<03:23, 13.83it/s] Update steps: 72%|█████████████████▎ | 7191/10000 [13:20<03:23, 13.82it/s] Update steps: 72%|█████████████████▎ | 7193/10000 [13:20<03:24, 13.73it/s] Update steps: 72%|█████████████████▎ | 7195/10000 [13:20<03:22, 13.84it/s] Update steps: 72%|█████████████████▎ | 7197/10000 [13:20<03:23, 13.79it/s] Update steps: 72%|█████████████████▎ | 7199/10000 [13:20<03:24, 13.72it/s] Update steps: 72%|█████████████████▎ | 7201/10000 [13:20<03:24, 13.70it/s] Update steps: 72%|█████████████████▎ | 7203/10000 [13:21<03:22, 13.82it/s] Update steps: 72%|█████████████████▎ | 7205/10000 [13:21<03:22, 13.79it/s] Update steps: 72%|█████████████████▎ | 7207/10000 [13:21<03:21, 13.85it/s] Update steps: 72%|█████████████████▎ | 7209/10000 [13:21<03:21, 13.83it/s] Update steps: 72%|█████████████████▎ | 7211/10000 [13:21<03:21, 13.84it/s] Update steps: 72%|█████████████████▎ | 7213/10000 [13:21<03:20, 13.87it/s] Update steps: 72%|█████████████████▎ | 7215/10000 [13:21<03:19, 13.96it/s] Update steps: 72%|█████████████████▎ | 7217/10000 [13:22<03:18, 14.00it/s] Update steps: 72%|█████████████████▎ | 7219/10000 [13:22<03:19, 13.97it/s] Update steps: 72%|█████████████████▎ | 7221/10000 [13:22<03:19, 13.94it/s] Update steps: 72%|█████████████████▎ | 7223/10000 [13:22<03:20, 13.87it/s] Update steps: 72%|█████████████████▎ | 7225/10000 [13:22<03:24, 13.56it/s] Update steps: 72%|█████████████████▎ | 7227/10000 [13:22<03:23, 13.60it/s] Update steps: 72%|█████████████████▎ | 7229/10000 [13:22<03:24, 13.53it/s] Update steps: 72%|█████████████████▎ | 7231/10000 [13:23<03:23, 13.60it/s] Update steps: 72%|█████████████████▎ | 7233/10000 [13:23<03:21, 13.71it/s] Update steps: 72%|█████████████████▎ | 7235/10000 [13:23<03:20, 13.77it/s] Update steps: 72%|█████████████████▎ | 7237/10000 [13:23<03:20, 13.76it/s] Update steps: 72%|█████████████████▎ | 7239/10000 [13:23<03:18, 13.88it/s] Update steps: 72%|█████████████████▍ | 7241/10000 [13:23<03:17, 13.94it/s] Update steps: 72%|█████████████████▍ | 7243/10000 [13:23<03:17, 13.96it/s] Update steps: 72%|█████████████████▍ | 7245/10000 [13:24<03:18, 13.89it/s] Update steps: 72%|█████████████████▍ | 7247/10000 [13:24<03:17, 13.91it/s] Update steps: 72%|█████████████████▍ | 7249/10000 [13:24<03:18, 13.87it/s] Update steps: 73%|█████████████████▍ | 7251/10000 [13:24<03:18, 13.86it/s] Update steps: 73%|█████████████████▍ | 7253/10000 [13:24<03:18, 13.83it/s] Update steps: 73%|█████████████████▍ | 7255/10000 [13:24<03:18, 13.86it/s] Update steps: 73%|█████████████████▍ | 7257/10000 [13:24<03:17, 13.87it/s] Update steps: 73%|█████████████████▍ | 7259/10000 [13:25<03:17, 13.89it/s] Update steps: 73%|█████████████████▍ | 7261/10000 [13:25<03:16, 13.95it/s] Update steps: 73%|█████████████████▍ | 7263/10000 [13:25<03:16, 13.96it/s] Update steps: 73%|█████████████████▍ | 7265/10000 [13:25<03:15, 13.98it/s] Update steps: 73%|█████████████████▍ | 7267/10000 [13:25<03:15, 13.99it/s] Update steps: 73%|█████████████████▍ | 7269/10000 [13:25<03:16, 13.92it/s] Update steps: 73%|█████████████████▍ | 7271/10000 [13:25<03:15, 13.94it/s] Update steps: 73%|█████████████████▍ | 7273/10000 [13:26<03:15, 13.95it/s] Update steps: 73%|█████████████████▍ | 7275/10000 [13:26<03:14, 13.98it/s] Update steps: 73%|█████████████████▍ | 7277/10000 [13:26<03:15, 13.96it/s] Update steps: 73%|█████████████████▍ | 7279/10000 [13:26<03:14, 14.01it/s] Update steps: 73%|█████████████████▍ | 7281/10000 [13:26<03:14, 13.95it/s] Update steps: 73%|█████████████████▍ | 7283/10000 [13:26<03:15, 13.89it/s] Update steps: 73%|█████████████████▍ | 7285/10000 [13:26<03:18, 13.65it/s] Update steps: 73%|█████████████████▍ | 7287/10000 [13:27<03:18, 13.68it/s] Update steps: 73%|█████████████████▍ | 7289/10000 [13:27<03:18, 13.66it/s] Update steps: 73%|█████████████████▍ | 7291/10000 [13:27<03:17, 13.70it/s] Update steps: 73%|█████████████████▌ | 7293/10000 [13:27<03:16, 13.79it/s] Update steps: 73%|█████████████████▌ | 7295/10000 [13:27<03:15, 13.85it/s] Update steps: 73%|█████████████████▌ | 7297/10000 [13:27<03:15, 13.82it/s] Update steps: 73%|█████████████████▌ | 7299/10000 [13:27<03:15, 13.82it/s] Update steps: 73%|█████████████████▌ | 7301/10000 [13:28<03:14, 13.90it/s] Update steps: 73%|█████████████████▌ | 7303/10000 [13:28<03:13, 13.93it/s] Update steps: 73%|█████████████████▌ | 7305/10000 [13:28<03:13, 13.95it/s] Update steps: 73%|█████████████████▌ | 7307/10000 [13:28<03:14, 13.83it/s] Update steps: 73%|█████████████████▌ | 7309/10000 [13:28<03:16, 13.72it/s] Update steps: 73%|█████████████████▌ | 7311/10000 [13:28<03:15, 13.74it/s] Update steps: 73%|█████████████████▌ | 7313/10000 [13:29<03:18, 13.50it/s] Update steps: 73%|█████████████████▌ | 7315/10000 [13:29<03:16, 13.66it/s] Update steps: 73%|█████████████████▌ | 7317/10000 [13:29<03:16, 13.66it/s] Update steps: 73%|█████████████████▌ | 7319/10000 [13:29<03:14, 13.77it/s] Update steps: 73%|█████████████████▌ | 7321/10000 [13:29<03:14, 13.77it/s] Update steps: 73%|█████████████████▌ | 7323/10000 [13:29<03:13, 13.84it/s] Update steps: 73%|█████████████████▌ | 7325/10000 [13:29<03:12, 13.89it/s] Update steps: 73%|█████████████████▌ | 7327/10000 [13:30<03:12, 13.91it/s] Update steps: 73%|█████████████████▌ | 7329/10000 [13:30<03:11, 13.96it/s] Update steps: 73%|█████████████████▌ | 7331/10000 [13:30<03:11, 13.94it/s] Update steps: 73%|█████████████████▌ | 7333/10000 [13:30<03:15, 13.63it/s] Update steps: 73%|█████████████████▌ | 7335/10000 [13:30<03:16, 13.55it/s] Update steps: 73%|█████████████████▌ | 7337/10000 [13:30<03:25, 12.98it/s] Update steps: 73%|█████████████████▌ | 7339/10000 [13:30<03:20, 13.29it/s] Update steps: 73%|█████████████████▌ | 7341/10000 [13:31<03:18, 13.41it/s] Update steps: 73%|█████████████████▌ | 7343/10000 [13:31<03:16, 13.51it/s] Update steps: 73%|█████████████████▋ | 7345/10000 [13:31<03:14, 13.68it/s] Update steps: 73%|█████████████████▋ | 7347/10000 [13:31<03:15, 13.56it/s] Update steps: 73%|█████████████████▋ | 7349/10000 [13:31<03:14, 13.66it/s] Update steps: 74%|█████████████████▋ | 7351/10000 [13:31<03:12, 13.79it/s] Update steps: 74%|█████████████████▋ | 7353/10000 [13:31<03:10, 13.87it/s] Update steps: 74%|█████████████████▋ | 7355/10000 [13:32<03:11, 13.81it/s] Update steps: 74%|█████████████████▋ | 7357/10000 [13:32<03:11, 13.80it/s] Update steps: 74%|█████████████████▋ | 7359/10000 [13:32<03:11, 13.76it/s] Update steps: 74%|█████████████████▋ | 7361/10000 [13:32<03:11, 13.75it/s] Update steps: 74%|█████████████████▋ | 7363/10000 [13:32<03:10, 13.86it/s] Update steps: 74%|█████████████████▋ | 7365/10000 [13:32<03:08, 13.95it/s] Update steps: 74%|█████████████████▋ | 7367/10000 [13:32<03:09, 13.91it/s] Update steps: 74%|█████████████████▋ | 7369/10000 [13:33<03:08, 13.95it/s] Update steps: 74%|█████████████████▋ | 7371/10000 [13:33<03:08, 13.94it/s] Update steps: 74%|█████████████████▋ | 7373/10000 [13:33<03:09, 13.84it/s] Update steps: 74%|█████████████████▋ | 7375/10000 [13:33<03:09, 13.86it/s] Update steps: 74%|█████████████████▋ | 7377/10000 [13:33<03:10, 13.76it/s] Update steps: 74%|█████████████████▋ | 7379/10000 [13:33<03:09, 13.86it/s] Update steps: 74%|█████████████████▋ | 7381/10000 [13:33<03:08, 13.90it/s] Update steps: 74%|█████████████████▋ | 7383/10000 [13:34<03:08, 13.90it/s] Update steps: 74%|█████████████████▋ | 7385/10000 [13:34<03:07, 13.93it/s] Update steps: 74%|█████████████████▋ | 7387/10000 [13:34<03:07, 13.96it/s] Update steps: 74%|█████████████████▋ | 7389/10000 [13:34<03:06, 13.98it/s] Update steps: 74%|█████████████████▋ | 7391/10000 [13:34<03:06, 13.99it/s] Update steps: 74%|█████████████████▋ | 7393/10000 [13:34<03:07, 13.93it/s] Update steps: 74%|█████████████████▋ | 7395/10000 [13:34<03:09, 13.76it/s] Update steps: 74%|█████████████████▊ | 7397/10000 [13:35<03:09, 13.71it/s] Update steps: 74%|█████████████████▊ | 7399/10000 [13:35<03:08, 13.82it/s] Update steps: 74%|█████████████████▊ | 7401/10000 [13:35<03:08, 13.77it/s] Update steps: 74%|█████████████████▊ | 7403/10000 [13:35<03:07, 13.85it/s] Update steps: 74%|█████████████████▊ | 7405/10000 [13:35<03:07, 13.83it/s] Update steps: 74%|█████████████████▊ | 7407/10000 [13:35<03:09, 13.65it/s] Update steps: 74%|█████████████████▊ | 7409/10000 [13:35<03:08, 13.73it/s] Update steps: 74%|█████████████████▊ | 7411/10000 [13:36<03:08, 13.76it/s] Update steps: 74%|█████████████████▊ | 7413/10000 [13:36<03:08, 13.75it/s] Update steps: 74%|█████████████████▊ | 7415/10000 [13:36<03:06, 13.83it/s] Update steps: 74%|█████████████████▊ | 7417/10000 [13:36<03:07, 13.75it/s] Update steps: 74%|█████████████████▊ | 7419/10000 [13:36<03:07, 13.77it/s] Update steps: 74%|█████████████████▊ | 7421/10000 [13:36<03:05, 13.89it/s] Update steps: 74%|█████████████████▊ | 7423/10000 [13:36<03:05, 13.92it/s] Update steps: 74%|█████████████████▊ | 7425/10000 [13:37<03:05, 13.90it/s] Update steps: 74%|█████████████████▊ | 7427/10000 [13:37<03:06, 13.78it/s] Update steps: 74%|█████████████████▊ | 7429/10000 [13:37<03:07, 13.74it/s] Update steps: 74%|█████████████████▊ | 7431/10000 [13:37<03:06, 13.78it/s] Update steps: 74%|█████████████████▊ | 7433/10000 [13:37<03:06, 13.80it/s] Update steps: 74%|█████████████████▊ | 7435/10000 [13:37<03:06, 13.76it/s] Update steps: 74%|█████████████████▊ | 7437/10000 [13:37<03:05, 13.85it/s] Update steps: 74%|█████████████████▊ | 7439/10000 [13:38<03:04, 13.91it/s] Update steps: 74%|█████████████████▊ | 7441/10000 [13:38<03:03, 13.97it/s] Update steps: 74%|█████████████████▊ | 7443/10000 [13:38<03:03, 13.95it/s] Update steps: 74%|█████████████████▊ | 7445/10000 [13:38<03:03, 13.95it/s] Update steps: 74%|█████████████████▊ | 7447/10000 [13:38<03:05, 13.79it/s] Update steps: 74%|█████████████████▉ | 7449/10000 [13:38<03:03, 13.90it/s] Update steps: 75%|█████████████████▉ | 7451/10000 [13:39<03:03, 13.90it/s] Update steps: 75%|█████████████████▉ | 7453/10000 [13:39<03:03, 13.89it/s] Update steps: 75%|█████████████████▉ | 7455/10000 [13:39<03:04, 13.78it/s] Update steps: 75%|█████████████████▉ | 7457/10000 [13:39<03:04, 13.80it/s] Update steps: 75%|█████████████████▉ | 7459/10000 [13:39<03:05, 13.73it/s] Update steps: 75%|█████████████████▉ | 7461/10000 [13:39<03:05, 13.72it/s] Update steps: 75%|█████████████████▉ | 7463/10000 [13:39<03:05, 13.67it/s] Update steps: 75%|█████████████████▉ | 7465/10000 [13:40<03:03, 13.79it/s] Update steps: 75%|█████████████████▉ | 7467/10000 [13:40<03:04, 13.70it/s] Update steps: 75%|█████████████████▉ | 7469/10000 [13:40<03:07, 13.50it/s] Update steps: 75%|█████████████████▉ | 7471/10000 [13:40<03:06, 13.59it/s] Update steps: 75%|█████████████████▉ | 7473/10000 [13:40<03:03, 13.76it/s] Update steps: 75%|█████████████████▉ | 7475/10000 [13:40<03:02, 13.85it/s] Update steps: 75%|█████████████████▉ | 7477/10000 [13:40<03:01, 13.92it/s] Update steps: 75%|█████████████████▉ | 7479/10000 [13:41<03:01, 13.88it/s] Update steps: 75%|█████████████████▉ | 7481/10000 [13:41<03:01, 13.91it/s] Update steps: 75%|█████████████████▉ | 7483/10000 [13:41<03:02, 13.78it/s] Update steps: 75%|█████████████████▉ | 7485/10000 [13:41<03:03, 13.70it/s] Update steps: 75%|█████████████████▉ | 7487/10000 [13:41<03:03, 13.69it/s] Update steps: 75%|█████████████████▉ | 7489/10000 [13:41<03:02, 13.79it/s] Update steps: 75%|█████████████████▉ | 7491/10000 [13:41<03:01, 13.83it/s] Update steps: 75%|█████████████████▉ | 7493/10000 [13:42<03:01, 13.80it/s] Update steps: 75%|█████████████████▉ | 7495/10000 [13:42<03:01, 13.83it/s] Update steps: 75%|█████████████████▉ | 7497/10000 [13:42<03:00, 13.87it/s] Update steps: 75%|█████████████████▉ | 7499/10000 [13:42<03:03, 13.66it/s] Update steps: 75%|██████████████████ | 7501/10000 [13:42<03:02, 13.69it/s] Update steps: 75%|██████████████████ | 7503/10000 [13:42<03:01, 13.75it/s] Update steps: 75%|██████████████████ | 7505/10000 [13:42<03:00, 13.86it/s] Update steps: 75%|██████████████████ | 7507/10000 [13:43<03:00, 13.79it/s] Update steps: 75%|██████████████████ | 7509/10000 [13:43<03:01, 13.75it/s] Update steps: 75%|██████████████████ | 7511/10000 [13:43<03:00, 13.82it/s] Update steps: 75%|██████████████████ | 7513/10000 [13:43<02:58, 13.90it/s] Update steps: 75%|██████████████████ | 7515/10000 [13:43<02:58, 13.91it/s] Update steps: 75%|██████████████████ | 7517/10000 [13:43<03:00, 13.76it/s] Update steps: 75%|██████████████████ | 7519/10000 [13:43<03:02, 13.63it/s] Update steps: 75%|██████████████████ | 7521/10000 [13:44<03:02, 13.60it/s] Update steps: 75%|██████████████████ | 7523/10000 [13:44<03:02, 13.59it/s] Update steps: 75%|██████████████████ | 7525/10000 [13:44<03:00, 13.73it/s] Update steps: 75%|██████████████████ | 7527/10000 [13:44<02:58, 13.82it/s] Update steps: 75%|██████████████████ | 7529/10000 [13:44<02:58, 13.84it/s] Update steps: 75%|██████████████████ | 7531/10000 [13:44<03:00, 13.67it/s] Update steps: 75%|██████████████████ | 7533/10000 [13:44<03:00, 13.65it/s] Update steps: 75%|██████████████████ | 7535/10000 [13:45<03:01, 13.57it/s] Update steps: 75%|██████████████████ | 7537/10000 [13:45<02:59, 13.69it/s] Update steps: 75%|██████████████████ | 7539/10000 [13:45<02:58, 13.75it/s] Update steps: 75%|██████████████████ | 7541/10000 [13:45<02:58, 13.80it/s] Update steps: 75%|██████████████████ | 7543/10000 [13:45<02:59, 13.66it/s] Update steps: 75%|██████████████████ | 7545/10000 [13:45<02:59, 13.71it/s] Update steps: 75%|██████████████████ | 7547/10000 [13:45<02:59, 13.69it/s] Update steps: 75%|██████████████████ | 7549/10000 [13:46<02:56, 13.85it/s] Update steps: 76%|██████████████████ | 7551/10000 [13:46<02:57, 13.83it/s] Update steps: 76%|██████████████████▏ | 7553/10000 [13:46<02:56, 13.87it/s] Update steps: 76%|██████████████████▏ | 7555/10000 [13:46<02:55, 13.92it/s] Update steps: 76%|██████████████████▏ | 7557/10000 [13:46<02:54, 13.96it/s] Update steps: 76%|██████████████████▏ | 7559/10000 [13:46<02:56, 13.84it/s] Update steps: 76%|██████████████████▏ | 7561/10000 [13:46<02:55, 13.92it/s] Update steps: 76%|██████████████████▏ | 7563/10000 [13:47<02:55, 13.87it/s] Update steps: 76%|██████████████████▏ | 7565/10000 [13:47<02:55, 13.90it/s] Update steps: 76%|██████████████████▏ | 7567/10000 [13:47<02:56, 13.75it/s] Update steps: 76%|██████████████████▏ | 7569/10000 [13:47<02:57, 13.72it/s] Update steps: 76%|██████████████████▏ | 7571/10000 [13:47<02:55, 13.82it/s] Update steps: 76%|██████████████████▏ | 7573/10000 [13:47<02:57, 13.65it/s] Update steps: 76%|██████████████████▏ | 7575/10000 [13:48<02:57, 13.66it/s] Update steps: 76%|██████████████████▏ | 7577/10000 [13:48<02:56, 13.74it/s] Update steps: 76%|██████████████████▏ | 7579/10000 [13:48<02:56, 13.71it/s] Update steps: 76%|██████████████████▏ | 7581/10000 [13:48<02:58, 13.57it/s] Update steps: 76%|██████████████████▏ | 7583/10000 [13:48<02:59, 13.46it/s] Update steps: 76%|██████████████████▏ | 7585/10000 [13:48<02:56, 13.65it/s] Update steps: 76%|██████████████████▏ | 7587/10000 [13:48<02:56, 13.71it/s] Update steps: 76%|██████████████████▏ | 7589/10000 [13:49<02:55, 13.76it/s] Update steps: 76%|██████████████████▏ | 7591/10000 [13:49<02:55, 13.70it/s] Update steps: 76%|██████████████████▏ | 7593/10000 [13:49<02:55, 13.73it/s] Update steps: 76%|██████████████████▏ | 7595/10000 [13:49<02:54, 13.80it/s] Update steps: 76%|██████████████████▏ | 7597/10000 [13:49<02:53, 13.85it/s] Update steps: 76%|██████████████████▏ | 7599/10000 [13:49<02:54, 13.78it/s] Update steps: 76%|██████████████████▏ | 7601/10000 [13:49<02:55, 13.71it/s] Update steps: 76%|██████████████████▏ | 7603/10000 [13:50<02:57, 13.54it/s] Update steps: 76%|██████████████████▎ | 7605/10000 [13:50<02:55, 13.68it/s] Update steps: 76%|██████████████████▎ | 7607/10000 [13:50<02:55, 13.65it/s] Update steps: 76%|██████████████████▎ | 7609/10000 [13:50<02:53, 13.76it/s] Update steps: 76%|██████████████████▎ | 7611/10000 [13:50<02:52, 13.87it/s] Update steps: 76%|██████████████████▎ | 7613/10000 [13:50<02:51, 13.93it/s] Update steps: 76%|██████████████████▎ | 7615/10000 [13:50<02:50, 13.98it/s] Update steps: 76%|██████████████████▎ | 7617/10000 [13:51<02:49, 14.04it/s] Update steps: 76%|██████████████████▎ | 7619/10000 [13:51<02:49, 14.01it/s] Update steps: 76%|██████████████████▎ | 7621/10000 [13:51<02:49, 14.05it/s] Update steps: 76%|██████████████████▎ | 7623/10000 [13:51<02:50, 13.97it/s] Update steps: 76%|██████████████████▎ | 7625/10000 [13:51<02:49, 13.99it/s] Update steps: 76%|██████████████████▎ | 7627/10000 [13:51<02:50, 13.90it/s] Update steps: 76%|██████████████████▎ | 7629/10000 [13:51<02:51, 13.85it/s] Update steps: 76%|██████████████████▎ | 7631/10000 [13:52<02:50, 13.92it/s] Update steps: 76%|██████████████████▎ | 7633/10000 [13:52<02:50, 13.89it/s] Update steps: 76%|██████████████████▎ | 7635/10000 [13:52<02:49, 13.95it/s] Update steps: 76%|██████████████████▎ | 7637/10000 [13:52<02:50, 13.89it/s] Update steps: 76%|██████████████████▎ | 7639/10000 [13:52<02:48, 13.98it/s] Update steps: 76%|██████████████████▎ | 7641/10000 [13:52<02:48, 14.01it/s] Update steps: 76%|██████████████████▎ | 7643/10000 [13:52<02:48, 13.97it/s] Update steps: 76%|██████████████████▎ | 7645/10000 [13:53<02:50, 13.83it/s] Update steps: 76%|██████████████████▎ | 7647/10000 [13:53<02:50, 13.80it/s] Update steps: 76%|██████████████████▎ | 7649/10000 [13:53<02:49, 13.91it/s] Update steps: 77%|██████████████████▎ | 7651/10000 [13:53<02:49, 13.83it/s] Update steps: 77%|██████████████████▎ | 7653/10000 [13:53<02:48, 13.92it/s] Update steps: 77%|██████████████████▎ | 7655/10000 [13:53<02:48, 13.93it/s] Update steps: 77%|██████████████████▍ | 7657/10000 [13:53<02:49, 13.79it/s] Update steps: 77%|██████████████████▍ | 7659/10000 [13:54<02:49, 13.78it/s] Update steps: 77%|██████████████████▍ | 7661/10000 [13:54<02:48, 13.88it/s] Update steps: 77%|██████████████████▍ | 7663/10000 [13:54<02:48, 13.85it/s] Update steps: 77%|██████████████████▍ | 7665/10000 [13:54<02:47, 13.94it/s] Update steps: 77%|██████████████████▍ | 7667/10000 [13:54<02:47, 13.96it/s] Update steps: 77%|██████████████████▍ | 7669/10000 [13:54<02:48, 13.80it/s] Update steps: 77%|██████████████████▍ | 7671/10000 [13:54<02:48, 13.83it/s] Update steps: 77%|██████████████████▍ | 7673/10000 [13:55<02:49, 13.73it/s] Update steps: 77%|██████████████████▍ | 7675/10000 [13:55<02:48, 13.79it/s] Update steps: 77%|██████████████████▍ | 7677/10000 [13:55<02:47, 13.88it/s] Update steps: 77%|██████████████████▍ | 7679/10000 [13:55<02:47, 13.85it/s] Update steps: 77%|██████████████████▍ | 7681/10000 [13:55<02:46, 13.90it/s] Update steps: 77%|██████████████████▍ | 7683/10000 [13:55<02:46, 13.95it/s] Update steps: 77%|██████████████████▍ | 7685/10000 [13:55<02:45, 13.99it/s] Update steps: 77%|██████████████████▍ | 7687/10000 [13:56<02:44, 14.04it/s] Update steps: 77%|██████████████████▍ | 7689/10000 [13:56<02:44, 14.06it/s] Update steps: 77%|██████████████████▍ | 7691/10000 [13:56<02:44, 14.06it/s] Update steps: 77%|██████████████████▍ | 7693/10000 [13:56<02:44, 13.99it/s] Update steps: 77%|██████████████████▍ | 7695/10000 [13:56<02:46, 13.88it/s] Update steps: 77%|██████████████████▍ | 7697/10000 [13:56<02:46, 13.87it/s] Update steps: 77%|██████████████████▍ | 7699/10000 [13:56<02:45, 13.92it/s] Update steps: 77%|██████████████████▍ | 7701/10000 [13:57<02:45, 13.93it/s] Update steps: 77%|██████████████████▍ | 7703/10000 [13:57<02:46, 13.81it/s] Update steps: 77%|██████████████████▍ | 7705/10000 [13:57<02:46, 13.78it/s] Update steps: 77%|██████████████████▍ | 7707/10000 [13:57<02:46, 13.79it/s] Update steps: 77%|██████████████████▌ | 7709/10000 [13:57<02:45, 13.82it/s] Update steps: 77%|██████████████████▌ | 7711/10000 [13:57<02:44, 13.91it/s] Update steps: 77%|██████████████████▌ | 7713/10000 [13:57<02:43, 13.96it/s] Update steps: 77%|██████████████████▌ | 7715/10000 [13:58<02:43, 13.94it/s] Update steps: 77%|██████████████████▌ | 7717/10000 [13:58<02:44, 13.85it/s] Update steps: 77%|██████████████████▌ | 7719/10000 [13:58<02:44, 13.84it/s] Update steps: 77%|██████████████████▌ | 7721/10000 [13:58<02:43, 13.91it/s] Update steps: 77%|██████████████████▌ | 7723/10000 [13:58<02:43, 13.90it/s] Update steps: 77%|██████████████████▌ | 7725/10000 [13:58<02:43, 13.91it/s] Update steps: 77%|██████████████████▌ | 7727/10000 [13:58<02:45, 13.73it/s] Update steps: 77%|██████████████████▌ | 7729/10000 [13:59<02:43, 13.89it/s] Update steps: 77%|██████████████████▌ | 7731/10000 [13:59<02:42, 13.93it/s] Update steps: 77%|██████████████████▌ | 7733/10000 [13:59<02:42, 13.94it/s] Update steps: 77%|██████████████████▌ | 7735/10000 [13:59<02:42, 13.98it/s] Update steps: 77%|██████████████████▌ | 7737/10000 [13:59<02:41, 13.99it/s] Update steps: 77%|██████████████████▌ | 7739/10000 [13:59<02:41, 14.02it/s] Update steps: 77%|██████████████████▌ | 7741/10000 [13:59<02:41, 13.96it/s] Update steps: 77%|██████████████████▌ | 7743/10000 [14:00<02:41, 13.99it/s] Update steps: 77%|██████████████████▌ | 7745/10000 [14:00<02:43, 13.80it/s] Update steps: 77%|██████████████████▌ | 7747/10000 [14:00<02:45, 13.63it/s] Update steps: 77%|██████████████████▌ | 7749/10000 [14:00<02:44, 13.68it/s] Update steps: 78%|██████████████████▌ | 7751/10000 [14:00<02:43, 13.76it/s] Update steps: 78%|██████████████████▌ | 7753/10000 [14:00<02:42, 13.85it/s] Update steps: 78%|██████████████████▌ | 7755/10000 [14:01<02:41, 13.87it/s] Update steps: 78%|██████████████████▌ | 7757/10000 [14:01<02:41, 13.93it/s] Update steps: 78%|██████████████████▌ | 7759/10000 [14:01<02:41, 13.86it/s] Update steps: 78%|██████████████████▋ | 7761/10000 [14:01<02:41, 13.88it/s] Update steps: 78%|██████████████████▋ | 7763/10000 [14:01<02:40, 13.91it/s] Update steps: 78%|██████████████████▋ | 7765/10000 [14:01<02:40, 13.92it/s] Update steps: 78%|██████████████████▋ | 7767/10000 [14:01<02:40, 13.91it/s] Update steps: 78%|██████████████████▋ | 7769/10000 [14:02<02:40, 13.87it/s] Update steps: 78%|██████████████████▋ | 7771/10000 [14:02<02:40, 13.88it/s] Update steps: 78%|██████████████████▋ | 7773/10000 [14:02<02:40, 13.86it/s] Update steps: 78%|██████████████████▋ | 7775/10000 [14:02<02:41, 13.76it/s] Update steps: 78%|██████████████████▋ | 7777/10000 [14:02<02:42, 13.71it/s] Update steps: 78%|██████████████████▋ | 7779/10000 [14:02<02:41, 13.72it/s] Update steps: 78%|██████████████████▋ | 7781/10000 [14:02<02:40, 13.79it/s] Update steps: 78%|██████████████████▋ | 7783/10000 [14:03<02:39, 13.87it/s] Update steps: 78%|██████████████████▋ | 7785/10000 [14:03<02:40, 13.83it/s] Update steps: 78%|██████████████████▋ | 7787/10000 [14:03<02:40, 13.79it/s] Update steps: 78%|██████████████████▋ | 7789/10000 [14:03<02:42, 13.64it/s] Update steps: 78%|██████████████████▋ | 7791/10000 [14:03<02:41, 13.66it/s] Update steps: 78%|██████████████████▋ | 7793/10000 [14:03<02:42, 13.56it/s] Update steps: 78%|██████████████████▋ | 7795/10000 [14:03<02:45, 13.36it/s] Update steps: 78%|██████████████████▋ | 7797/10000 [14:04<02:43, 13.44it/s] Update steps: 78%|██████████████████▋ | 7799/10000 [14:04<02:42, 13.58it/s] Update steps: 78%|██████████████████▋ | 7801/10000 [14:04<02:40, 13.70it/s] Update steps: 78%|██████████████████▋ | 7803/10000 [14:04<02:39, 13.73it/s] Update steps: 78%|██████████████████▋ | 7805/10000 [14:04<02:41, 13.57it/s] Update steps: 78%|██████████████████▋ | 7807/10000 [14:04<02:41, 13.59it/s] Update steps: 78%|██████████████████▋ | 7809/10000 [14:04<02:40, 13.69it/s] Update steps: 78%|██████████████████▋ | 7811/10000 [14:05<02:38, 13.79it/s] Update steps: 78%|██████████████████▊ | 7813/10000 [14:05<02:38, 13.76it/s] Update steps: 78%|██████████████████▊ | 7815/10000 [14:05<02:38, 13.78it/s] Update steps: 78%|██████████████████▊ | 7817/10000 [14:05<02:39, 13.72it/s] Update steps: 78%|██████████████████▊ | 7819/10000 [14:05<02:37, 13.83it/s] Update steps: 78%|██████████████████▊ | 7821/10000 [14:05<02:38, 13.79it/s] Update steps: 78%|██████████████████▊ | 7823/10000 [14:05<02:36, 13.88it/s] Update steps: 78%|██████████████████▊ | 7825/10000 [14:06<02:36, 13.94it/s] Update steps: 78%|██████████████████▊ | 7827/10000 [14:06<02:35, 13.95it/s] Update steps: 78%|██████████████████▊ | 7829/10000 [14:06<02:37, 13.82it/s] Update steps: 78%|██████████████████▊ | 7831/10000 [14:06<02:37, 13.77it/s] Update steps: 78%|██████████████████▊ | 7833/10000 [14:06<02:36, 13.87it/s] Update steps: 78%|██████████████████▊ | 7835/10000 [14:06<02:35, 13.90it/s] Update steps: 78%|██████████████████▊ | 7837/10000 [14:06<02:34, 13.99it/s] Update steps: 78%|██████████████████▊ | 7839/10000 [14:07<02:35, 13.86it/s] Update steps: 78%|██████████████████▊ | 7841/10000 [14:07<02:35, 13.87it/s] Update steps: 78%|██████████████████▊ | 7843/10000 [14:07<02:34, 13.94it/s] Update steps: 78%|██████████████████▊ | 7845/10000 [14:07<02:34, 13.96it/s] Update steps: 78%|██████████████████▊ | 7847/10000 [14:07<02:34, 13.95it/s] Update steps: 78%|██████████████████▊ | 7849/10000 [14:07<02:36, 13.78it/s] Update steps: 79%|██████████████████▊ | 7851/10000 [14:07<02:36, 13.71it/s] Update steps: 79%|██████████████████▊ | 7853/10000 [14:08<02:36, 13.72it/s] Update steps: 79%|██████████████████▊ | 7855/10000 [14:08<02:36, 13.73it/s] Update steps: 79%|██████████████████▊ | 7857/10000 [14:08<02:34, 13.84it/s] Update steps: 79%|██████████████████▊ | 7859/10000 [14:08<02:34, 13.82it/s] Update steps: 79%|██████████████████▊ | 7861/10000 [14:08<02:33, 13.90it/s] Update steps: 79%|██████████████████▊ | 7863/10000 [14:08<02:35, 13.74it/s] Update steps: 79%|██████████████████▉ | 7865/10000 [14:08<02:34, 13.80it/s] Update steps: 79%|██████████████████▉ | 7867/10000 [14:09<02:33, 13.85it/s] Update steps: 79%|██████████████████▉ | 7869/10000 [14:09<02:33, 13.91it/s] Update steps: 79%|██████████████████▉ | 7871/10000 [14:09<02:32, 13.95it/s] Update steps: 79%|██████████████████▉ | 7873/10000 [14:09<02:33, 13.88it/s] Update steps: 79%|██████████████████▉ | 7875/10000 [14:09<02:33, 13.89it/s] Update steps: 79%|██████████████████▉ | 7877/10000 [14:09<02:32, 13.93it/s] Update steps: 79%|██████████████████▉ | 7879/10000 [14:09<02:32, 13.89it/s] Update steps: 79%|██████████████████▉ | 7881/10000 [14:10<02:31, 13.96it/s] Update steps: 79%|██████████████████▉ | 7883/10000 [14:10<02:32, 13.93it/s] Update steps: 79%|██████████████████▉ | 7885/10000 [14:10<02:33, 13.82it/s] Update steps: 79%|██████████████████▉ | 7887/10000 [14:10<02:33, 13.76it/s] Update steps: 79%|██████████████████▉ | 7889/10000 [14:10<02:32, 13.85it/s] Update steps: 79%|██████████████████▉ | 7891/10000 [14:10<02:33, 13.78it/s] Update steps: 79%|██████████████████▉ | 7893/10000 [14:10<02:32, 13.79it/s] Update steps: 79%|██████████████████▉ | 7895/10000 [14:11<02:31, 13.87it/s] Update steps: 79%|██████████████████▉ | 7897/10000 [14:11<02:33, 13.69it/s] Update steps: 79%|██████████████████▉ | 7899/10000 [14:11<02:35, 13.53it/s] Update steps: 79%|██████████████████▉ | 7901/10000 [14:11<02:34, 13.57it/s] Update steps: 79%|██████████████████▉ | 7903/10000 [14:11<02:34, 13.60it/s] Update steps: 79%|██████████████████▉ | 7905/10000 [14:11<02:33, 13.63it/s] Update steps: 79%|██████████████████▉ | 7907/10000 [14:12<02:32, 13.70it/s] Update steps: 79%|██████████████████▉ | 7909/10000 [14:12<02:31, 13.84it/s] Update steps: 79%|██████████████████▉ | 7911/10000 [14:12<02:31, 13.83it/s] Update steps: 79%|██████████████████▉ | 7913/10000 [14:12<02:30, 13.89it/s] Update steps: 79%|██████████████████▉ | 7915/10000 [14:12<02:30, 13.81it/s] Update steps: 79%|███████████████████ | 7917/10000 [14:12<02:30, 13.86it/s] Update steps: 79%|███████████████████ | 7919/10000 [14:12<02:29, 13.90it/s] Update steps: 79%|███████████████████ | 7921/10000 [14:13<02:28, 13.95it/s] Update steps: 79%|███████████████████ | 7923/10000 [14:13<02:29, 13.85it/s] Update steps: 79%|███████████████████ | 7925/10000 [14:13<02:29, 13.84it/s] Update steps: 79%|███████████████████ | 7927/10000 [14:13<02:28, 13.92it/s] Update steps: 79%|███████████████████ | 7929/10000 [14:13<02:28, 13.95it/s] Update steps: 79%|███████████████████ | 7931/10000 [14:13<02:29, 13.82it/s] Update steps: 79%|███████████████████ | 7933/10000 [14:13<02:30, 13.74it/s] Update steps: 79%|███████████████████ | 7935/10000 [14:14<02:29, 13.80it/s] Update steps: 79%|███████████████████ | 7937/10000 [14:14<02:29, 13.76it/s] Update steps: 79%|███████████████████ | 7939/10000 [14:14<02:29, 13.83it/s] Update steps: 79%|███████████████████ | 7941/10000 [14:14<02:29, 13.74it/s] Update steps: 79%|███████████████████ | 7943/10000 [14:14<02:30, 13.68it/s] Update steps: 79%|███████████████████ | 7945/10000 [14:14<02:30, 13.65it/s] Update steps: 79%|███████████████████ | 7947/10000 [14:14<02:29, 13.69it/s] Update steps: 79%|███████████████████ | 7949/10000 [14:15<02:29, 13.76it/s] Update steps: 80%|███████████████████ | 7951/10000 [14:15<02:27, 13.85it/s] Update steps: 80%|███████████████████ | 7953/10000 [14:15<02:26, 13.93it/s] Update steps: 80%|███████████████████ | 7955/10000 [14:15<02:27, 13.91it/s] Update steps: 80%|███████████████████ | 7957/10000 [14:15<02:27, 13.83it/s] Update steps: 80%|███████████████████ | 7959/10000 [14:15<02:27, 13.86it/s] Update steps: 80%|███████████████████ | 7961/10000 [14:15<02:27, 13.79it/s] Update steps: 80%|███████████████████ | 7963/10000 [14:16<02:26, 13.86it/s] Update steps: 80%|███████████████████ | 7965/10000 [14:16<02:26, 13.89it/s] Update steps: 80%|███████████████████ | 7967/10000 [14:16<02:25, 13.97it/s] Update steps: 80%|███████████████████▏ | 7969/10000 [14:16<02:25, 13.99it/s] Update steps: 80%|███████████████████▏ | 7971/10000 [14:16<02:24, 14.02it/s] Update steps: 80%|███████████████████▏ | 7973/10000 [14:16<02:24, 14.04it/s] Update steps: 80%|███████████████████▏ | 7975/10000 [14:16<02:23, 14.06it/s] Update steps: 80%|███████████████████▏ | 7977/10000 [14:17<02:23, 14.09it/s] Update steps: 80%|███████████████████▏ | 7979/10000 [14:17<02:23, 14.09it/s] Update steps: 80%|███████████████████▏ | 7981/10000 [14:17<02:23, 14.10it/s] Update steps: 80%|███████████████████▏ | 7983/10000 [14:17<02:24, 13.94it/s] Update steps: 80%|███████████████████▏ | 7985/10000 [14:17<02:23, 13.99it/s] Update steps: 80%|███████████████████▏ | 7987/10000 [14:17<02:24, 13.95it/s] Update steps: 80%|███████████████████▏ | 7989/10000 [14:17<02:24, 13.93it/s] Update steps: 80%|███████████████████▏ | 7991/10000 [14:18<02:24, 13.94it/s] Update steps: 80%|███████████████████▏ | 7993/10000 [14:18<02:24, 13.85it/s] Update steps: 80%|███████████████████▏ | 7995/10000 [14:18<02:25, 13.80it/s] Update steps: 80%|███████████████████▏ | 7997/10000 [14:18<02:24, 13.86it/s] Update steps: 80%|███████████████████▏ | 7999/10000 [14:18<02:24, 13.85it/s]2026-04-17 11:07:39.112 | INFO | __main__:main:490 - Performing evaluation at step 8000 +2026-04-17 11:07:39.117 | INFO | utils.eval:evaluate_model:21 - Loading validation data from 8 local validation files +2026-04-17 11:07:39.906 | INFO | utils.eval:evaluate_model:43 - Loaded validation dataset in 0.79 seconds +2026-04-17 11:07:39.907 | INFO | utils.eval:evaluate_model:60 - Eval set prepared in 0.79 seconds + Update steps: 80%|███████████████████▏ | 8000/10000 [14:30<02:24, 13.85it/s]2026-04-17 11:08:14.483 | INFO | __main__:main:505 - Eval loss at step 8000: 3.426166321686302, Eval perplexity: 30.758498242683302 + Update steps: 80%|█████████████████▌ | 8001/10000 [14:54<2:59:09, 5.38s/it] Update steps: 80%|█████████████████▌ | 8003/10000 [14:54<2:05:58, 3.79s/it] Update steps: 80%|█████████████████▌ | 8005/10000 [14:54<1:28:47, 2.67s/it] Update steps: 80%|█████████████████▌ | 8007/10000 [14:54<1:02:48, 1.89s/it] Update steps: 80%|███████████████████▏ | 8009/10000 [14:54<44:37, 1.34s/it] Update steps: 80%|███████████████████▏ | 8011/10000 [14:54<31:55, 1.04it/s] Update steps: 80%|███████████████████▏ | 8013/10000 [14:55<23:01, 1.44it/s] Update steps: 80%|███████████████████▏ | 8015/10000 [14:55<16:48, 1.97it/s] Update steps: 80%|███████████████████▏ | 8017/10000 [14:55<12:27, 2.65it/s] Update steps: 80%|███████████████████▏ | 8019/10000 [14:55<09:24, 3.51it/s] Update steps: 80%|███████████████████▎ | 8021/10000 [14:55<07:17, 4.52it/s] Update steps: 80%|███████████████████▎ | 8023/10000 [14:55<05:47, 5.68it/s] Update steps: 80%|███████████████████▎ | 8025/10000 [14:55<04:45, 6.93it/s] Update steps: 80%|███████████████████▎ | 8027/10000 [14:56<04:02, 8.12it/s] Update steps: 80%|███████████████████▎ | 8029/10000 [14:56<03:32, 9.26it/s] Update steps: 80%|███████████████████▎ | 8031/10000 [14:56<03:10, 10.31it/s] Update steps: 80%|███████████████████▎ | 8033/10000 [14:56<02:57, 11.11it/s] Update steps: 80%|███████████████████▎ | 8035/10000 [14:56<02:46, 11.79it/s] Update steps: 80%|███████████████████▎ | 8037/10000 [14:56<02:38, 12.39it/s] Update steps: 80%|███████████████████▎ | 8039/10000 [14:56<02:32, 12.86it/s] Update steps: 80%|███████████████████▎ | 8041/10000 [14:57<02:28, 13.22it/s] Update steps: 80%|███████████████████▎ | 8043/10000 [14:57<02:27, 13.30it/s] Update steps: 80%|███████████████████▎ | 8045/10000 [14:57<02:24, 13.52it/s] Update steps: 80%|███████████████████▎ | 8047/10000 [14:57<02:22, 13.68it/s] Update steps: 80%|███████████████████▎ | 8049/10000 [14:57<02:21, 13.78it/s] Update steps: 81%|███████████████████▎ | 8051/10000 [14:57<02:21, 13.77it/s] Update steps: 81%|███████████████████▎ | 8053/10000 [14:57<02:20, 13.86it/s] Update steps: 81%|███████████████████▎ | 8055/10000 [14:58<02:20, 13.89it/s] Update steps: 81%|███████████████████▎ | 8057/10000 [14:58<02:19, 13.95it/s] Update steps: 81%|███████████████████▎ | 8059/10000 [14:58<02:19, 13.88it/s] Update steps: 81%|███████████████████▎ | 8061/10000 [14:58<02:19, 13.89it/s] Update steps: 81%|███████████████████▎ | 8063/10000 [14:58<02:19, 13.91it/s] Update steps: 81%|███████████████████▎ | 8065/10000 [14:58<02:18, 13.94it/s] Update steps: 81%|███████████████████▎ | 8067/10000 [14:58<02:18, 13.99it/s] Update steps: 81%|███████████████████▎ | 8069/10000 [14:59<02:18, 13.93it/s] Update steps: 81%|███████████████████▎ | 8071/10000 [14:59<02:18, 13.89it/s] Update steps: 81%|███████████████████▍ | 8073/10000 [14:59<02:18, 13.94it/s] Update steps: 81%|███████████████████▍ | 8075/10000 [14:59<02:18, 13.94it/s] Update steps: 81%|███████████████████▍ | 8077/10000 [14:59<02:18, 13.87it/s] Update steps: 81%|███████████████████▍ | 8079/10000 [14:59<02:18, 13.84it/s] Update steps: 81%|███████████████████▍ | 8081/10000 [14:59<02:21, 13.52it/s] Update steps: 81%|███████████████████▍ | 8083/10000 [15:00<02:20, 13.65it/s] Update steps: 81%|███████████████████▍ | 8085/10000 [15:00<02:19, 13.70it/s] Update steps: 81%|███████████████████▍ | 8087/10000 [15:00<02:19, 13.72it/s] Update steps: 81%|███████████████████▍ | 8089/10000 [15:00<02:18, 13.78it/s] Update steps: 81%|███████████████████▍ | 8091/10000 [15:00<02:18, 13.80it/s] Update steps: 81%|███████████████████▍ | 8093/10000 [15:00<02:17, 13.82it/s] Update steps: 81%|███████████████████▍ | 8095/10000 [15:00<02:18, 13.73it/s] Update steps: 81%|███████████████████▍ | 8097/10000 [15:01<02:17, 13.81it/s] Update steps: 81%|███████████████████▍ | 8099/10000 [15:01<02:16, 13.88it/s] Update steps: 81%|███████████████████▍ | 8101/10000 [15:01<02:16, 13.90it/s] Update steps: 81%|███████████████████▍ | 8103/10000 [15:01<02:17, 13.77it/s] Update steps: 81%|███████████████████▍ | 8105/10000 [15:01<02:16, 13.85it/s] Update steps: 81%|███████████████████▍ | 8107/10000 [15:01<02:16, 13.89it/s] Update steps: 81%|███████████████████▍ | 8109/10000 [15:01<02:15, 13.91it/s] Update steps: 81%|███████████████████▍ | 8111/10000 [15:02<02:15, 13.95it/s] Update steps: 81%|███████████████████▍ | 8113/10000 [15:02<02:14, 14.04it/s] Update steps: 81%|███████████████████▍ | 8115/10000 [15:02<02:15, 13.96it/s] Update steps: 81%|███████████████████▍ | 8117/10000 [15:02<02:14, 13.95it/s] Update steps: 81%|███████████████████▍ | 8119/10000 [15:02<02:15, 13.91it/s] Update steps: 81%|███████████████████▍ | 8121/10000 [15:02<02:15, 13.83it/s] Update steps: 81%|███████████████████▍ | 8123/10000 [15:02<02:14, 13.91it/s] Update steps: 81%|███████████████████▌ | 8125/10000 [15:03<02:14, 13.92it/s] Update steps: 81%|███████████████████▌ | 8127/10000 [15:03<02:14, 13.97it/s] Update steps: 81%|███████████████████▌ | 8129/10000 [15:03<02:13, 13.96it/s] Update steps: 81%|███████████████████▌ | 8131/10000 [15:03<02:13, 14.01it/s] Update steps: 81%|███████████████████▌ | 8133/10000 [15:03<02:13, 14.00it/s] Update steps: 81%|███████████████████▌ | 8135/10000 [15:03<02:13, 13.99it/s] Update steps: 81%|███████████████████▌ | 8137/10000 [15:03<02:12, 14.01it/s] Update steps: 81%|███████████████████▌ | 8139/10000 [15:04<02:12, 14.04it/s] Update steps: 81%|███████████████████▌ | 8141/10000 [15:04<02:12, 14.01it/s] Update steps: 81%|███████████████████▌ | 8143/10000 [15:04<02:13, 13.96it/s] Update steps: 81%|███████████████████▌ | 8145/10000 [15:04<02:13, 13.95it/s] Update steps: 81%|███████████████████▌ | 8147/10000 [15:04<02:13, 13.93it/s] Update steps: 81%|███████████████████▌ | 8149/10000 [15:04<02:15, 13.69it/s] Update steps: 82%|███████████████████▌ | 8151/10000 [15:04<02:14, 13.76it/s] Update steps: 82%|███████████████████▌ | 8153/10000 [15:05<02:13, 13.81it/s] Update steps: 82%|███████████████████▌ | 8155/10000 [15:05<02:13, 13.85it/s] Update steps: 82%|███████████████████▌ | 8157/10000 [15:05<02:13, 13.83it/s] Update steps: 82%|███████████████████▌ | 8159/10000 [15:05<02:13, 13.76it/s] Update steps: 82%|███████████████████▌ | 8161/10000 [15:05<02:13, 13.77it/s] Update steps: 82%|███████████████████▌ | 8163/10000 [15:05<02:13, 13.75it/s] Update steps: 82%|███████████████████▌ | 8165/10000 [15:05<02:12, 13.83it/s] Update steps: 82%|███████████████████▌ | 8167/10000 [15:06<02:13, 13.78it/s] Update steps: 82%|███████████████████▌ | 8169/10000 [15:06<02:12, 13.86it/s] Update steps: 82%|███████████████████▌ | 8171/10000 [15:06<02:11, 13.91it/s] Update steps: 82%|███████████████████▌ | 8173/10000 [15:06<02:11, 13.88it/s] Update steps: 82%|███████████████████▌ | 8175/10000 [15:06<02:12, 13.79it/s] Update steps: 82%|███████████████████▌ | 8177/10000 [15:06<02:12, 13.80it/s] Update steps: 82%|███████████████████▋ | 8179/10000 [15:06<02:11, 13.85it/s] Update steps: 82%|███████████████████▋ | 8181/10000 [15:07<02:11, 13.85it/s] Update steps: 82%|███████████████████▋ | 8183/10000 [15:07<02:13, 13.56it/s] Update steps: 82%|███████████████████▋ | 8185/10000 [15:07<02:12, 13.66it/s] Update steps: 82%|███████████████████▋ | 8187/10000 [15:07<02:11, 13.78it/s] Update steps: 82%|███████████████████▋ | 8189/10000 [15:07<02:11, 13.81it/s] Update steps: 82%|███████████████████▋ | 8191/10000 [15:07<02:11, 13.80it/s] Update steps: 82%|███████████████████▋ | 8193/10000 [15:07<02:11, 13.70it/s] Update steps: 82%|███████████████████▋ | 8195/10000 [15:08<02:11, 13.73it/s] Update steps: 82%|███████████████████▋ | 8197/10000 [15:08<02:11, 13.73it/s] Update steps: 82%|███████████████████▋ | 8199/10000 [15:08<02:10, 13.76it/s] Update steps: 82%|███████████████████▋ | 8201/10000 [15:08<02:10, 13.81it/s] Update steps: 82%|███████████████████▋ | 8203/10000 [15:08<02:09, 13.87it/s] Update steps: 82%|███████████████████▋ | 8205/10000 [15:08<02:09, 13.91it/s] Update steps: 82%|███████████████████▋ | 8207/10000 [15:08<02:08, 13.94it/s] Update steps: 82%|███████████████████▋ | 8209/10000 [15:09<02:10, 13.71it/s] Update steps: 82%|███████████████████▋ | 8211/10000 [15:09<02:09, 13.77it/s] Update steps: 82%|███████████████████▋ | 8213/10000 [15:09<02:09, 13.79it/s] Update steps: 82%|███████████████████▋ | 8215/10000 [15:09<02:09, 13.81it/s] Update steps: 82%|███████████████████▋ | 8217/10000 [15:09<02:08, 13.85it/s] Update steps: 82%|███████████████████▋ | 8219/10000 [15:09<02:08, 13.82it/s] Update steps: 82%|███████████████████▋ | 8221/10000 [15:10<02:08, 13.87it/s] Update steps: 82%|███████████████████▋ | 8223/10000 [15:10<02:07, 13.91it/s] Update steps: 82%|███████████████████▋ | 8225/10000 [15:10<02:07, 13.94it/s] Update steps: 82%|███████████████████▋ | 8227/10000 [15:10<02:07, 13.95it/s] Update steps: 82%|███████████████████▋ | 8229/10000 [15:10<02:06, 13.98it/s] Update steps: 82%|███████████████████▊ | 8231/10000 [15:10<02:06, 13.98it/s] Update steps: 82%|███████████████████▊ | 8233/10000 [15:10<02:06, 14.01it/s] Update steps: 82%|███████████████████▊ | 8235/10000 [15:11<02:05, 14.03it/s] Update steps: 82%|███████████████████▊ | 8237/10000 [15:11<02:05, 14.02it/s] Update steps: 82%|███████████████████▊ | 8239/10000 [15:11<02:06, 13.97it/s] Update steps: 82%|███████████████████▊ | 8241/10000 [15:11<02:05, 13.99it/s] Update steps: 82%|███████████████████▊ | 8243/10000 [15:11<02:06, 13.84it/s] Update steps: 82%|███████████████████▊ | 8245/10000 [15:11<02:07, 13.72it/s] Update steps: 82%|███████████████████▊ | 8247/10000 [15:11<02:08, 13.69it/s] Update steps: 82%|███████████████████▊ | 8249/10000 [15:12<02:08, 13.66it/s] Update steps: 83%|███████████████████▊ | 8251/10000 [15:12<02:13, 13.05it/s] Update steps: 83%|███████████████████▊ | 8253/10000 [15:12<02:13, 13.13it/s] Update steps: 83%|███████████████████▊ | 8255/10000 [15:12<02:09, 13.45it/s] Update steps: 83%|███████████████████▊ | 8257/10000 [15:12<02:08, 13.53it/s] Update steps: 83%|███████████████████▊ | 8259/10000 [15:12<02:07, 13.67it/s] Update steps: 83%|███████████████████▊ | 8261/10000 [15:12<02:06, 13.72it/s] Update steps: 83%|███████████████████▊ | 8263/10000 [15:13<02:05, 13.81it/s] Update steps: 83%|███████████████████▊ | 8265/10000 [15:13<02:05, 13.86it/s] Update steps: 83%|███████████████████▊ | 8267/10000 [15:13<02:04, 13.88it/s] Update steps: 83%|███████████████████▊ | 8269/10000 [15:13<02:04, 13.87it/s] Update steps: 83%|███████████████████▊ | 8271/10000 [15:13<02:04, 13.94it/s] Update steps: 83%|███████████████████▊ | 8273/10000 [15:13<02:06, 13.67it/s] Update steps: 83%|███████████████████▊ | 8275/10000 [15:13<02:05, 13.74it/s] Update steps: 83%|███████████████████▊ | 8277/10000 [15:14<02:05, 13.72it/s] Update steps: 83%|███████████████████▊ | 8279/10000 [15:14<02:05, 13.75it/s] Update steps: 83%|███████████████████▊ | 8281/10000 [15:14<02:04, 13.76it/s] Update steps: 83%|███████████████████▉ | 8283/10000 [15:14<02:04, 13.74it/s] Update steps: 83%|███████████████████▉ | 8285/10000 [15:14<02:04, 13.76it/s] Update steps: 83%|███████████████████▉ | 8287/10000 [15:14<02:04, 13.74it/s] Update steps: 83%|███████████████████▉ | 8289/10000 [15:14<02:03, 13.83it/s] Update steps: 83%|███████████████████▉ | 8291/10000 [15:15<02:03, 13.79it/s] Update steps: 83%|███████████████████▉ | 8293/10000 [15:15<02:02, 13.88it/s] Update steps: 83%|███████████████████▉ | 8295/10000 [15:15<02:03, 13.85it/s] Update steps: 83%|███████████████████▉ | 8297/10000 [15:15<02:03, 13.74it/s] Update steps: 83%|███████████████████▉ | 8299/10000 [15:15<02:04, 13.69it/s] Update steps: 83%|███████████████████▉ | 8301/10000 [15:15<02:03, 13.71it/s] Update steps: 83%|███████████████████▉ | 8303/10000 [15:15<02:02, 13.82it/s] Update steps: 83%|███████████████████▉ | 8305/10000 [15:16<02:02, 13.88it/s] Update steps: 83%|███████████████████▉ | 8307/10000 [15:16<02:01, 13.93it/s] Update steps: 83%|███████████████████▉ | 8309/10000 [15:16<02:00, 13.99it/s] Update steps: 83%|███████████████████▉ | 8311/10000 [15:16<02:00, 14.00it/s] Update steps: 83%|███████████████████▉ | 8313/10000 [15:16<02:04, 13.50it/s] Update steps: 83%|███████████████████▉ | 8315/10000 [15:16<02:03, 13.66it/s] Update steps: 83%|███████████████████▉ | 8317/10000 [15:16<02:02, 13.71it/s] Update steps: 83%|███████████████████▉ | 8319/10000 [15:17<02:02, 13.78it/s] Update steps: 83%|███████████████████▉ | 8321/10000 [15:17<02:01, 13.84it/s] Update steps: 83%|███████████████████▉ | 8323/10000 [15:17<02:01, 13.81it/s] Update steps: 83%|███████████████████▉ | 8325/10000 [15:17<02:00, 13.88it/s] Update steps: 83%|███████████████████▉ | 8327/10000 [15:17<02:00, 13.89it/s] Update steps: 83%|███████████████████▉ | 8329/10000 [15:17<01:59, 13.96it/s] Update steps: 83%|███████████████████▉ | 8331/10000 [15:17<02:00, 13.84it/s] Update steps: 83%|███████████████████▉ | 8333/10000 [15:18<01:59, 13.95it/s] Update steps: 83%|████████████████████ | 8335/10000 [15:18<01:59, 13.94it/s] Update steps: 83%|████████████████████ | 8337/10000 [15:18<01:59, 13.87it/s] Update steps: 83%|████████████████████ | 8339/10000 [15:18<01:59, 13.86it/s] Update steps: 83%|████████████████████ | 8341/10000 [15:18<01:59, 13.88it/s] Update steps: 83%|████████████████████ | 8343/10000 [15:18<01:59, 13.90it/s] Update steps: 83%|████████████████████ | 8345/10000 [15:18<01:58, 13.92it/s] Update steps: 83%|████████████████████ | 8347/10000 [15:19<01:58, 13.90it/s] Update steps: 83%|████████████████████ | 8349/10000 [15:19<01:58, 13.92it/s] Update steps: 84%|████████████████████ | 8351/10000 [15:19<01:58, 13.89it/s] Update steps: 84%|████████████████████ | 8353/10000 [15:19<01:58, 13.90it/s] Update steps: 84%|████████████████████ | 8355/10000 [15:19<01:58, 13.84it/s] Update steps: 84%|████████████████████ | 8357/10000 [15:19<01:58, 13.86it/s] Update steps: 84%|████████████████████ | 8359/10000 [15:20<01:57, 13.92it/s] Update steps: 84%|████████████████████ | 8361/10000 [15:20<01:57, 13.97it/s] Update steps: 84%|████████████████████ | 8363/10000 [15:20<01:56, 13.99it/s] Update steps: 84%|████████████████████ | 8365/10000 [15:20<01:56, 14.02it/s] Update steps: 84%|████████████████████ | 8367/10000 [15:20<01:56, 14.03it/s] Update steps: 84%|████████████████████ | 8369/10000 [15:20<01:57, 13.87it/s] Update steps: 84%|████████████████████ | 8371/10000 [15:20<01:57, 13.88it/s] Update steps: 84%|████████████████████ | 8373/10000 [15:21<01:57, 13.80it/s] Update steps: 84%|████████████████████ | 8375/10000 [15:21<01:56, 13.90it/s] Update steps: 84%|████████████████████ | 8377/10000 [15:21<01:56, 13.92it/s] Update steps: 84%|████████████████████ | 8379/10000 [15:21<01:56, 13.97it/s] Update steps: 84%|████████████████████ | 8381/10000 [15:21<01:56, 13.95it/s] Update steps: 84%|████████████████████ | 8383/10000 [15:21<01:56, 13.89it/s] Update steps: 84%|████████████████████ | 8385/10000 [15:21<01:55, 13.96it/s] Update steps: 84%|████████████████████▏ | 8387/10000 [15:22<01:56, 13.87it/s] Update steps: 84%|████████████████████▏ | 8389/10000 [15:22<01:55, 13.91it/s] Update steps: 84%|████████████████████▏ | 8391/10000 [15:22<01:55, 13.89it/s] Update steps: 84%|████████████████████▏ | 8393/10000 [15:22<01:56, 13.84it/s] Update steps: 84%|████████████████████▏ | 8395/10000 [15:22<01:55, 13.91it/s] Update steps: 84%|████████████████████▏ | 8397/10000 [15:22<01:55, 13.85it/s] Update steps: 84%|████████████████████▏ | 8399/10000 [15:22<01:55, 13.91it/s] Update steps: 84%|████████████████████▏ | 8401/10000 [15:23<01:57, 13.64it/s] Update steps: 84%|████████████████████▏ | 8403/10000 [15:23<01:57, 13.57it/s] Update steps: 84%|████████████████████▏ | 8405/10000 [15:23<01:56, 13.68it/s] Update steps: 84%|████████████████████▏ | 8407/10000 [15:23<01:55, 13.76it/s] Update steps: 84%|████████████████████▏ | 8409/10000 [15:23<01:55, 13.76it/s] Update steps: 84%|████████████████████▏ | 8411/10000 [15:23<01:55, 13.74it/s] Update steps: 84%|████████████████████▏ | 8413/10000 [15:23<01:55, 13.80it/s] Update steps: 84%|████████████████████▏ | 8415/10000 [15:24<01:54, 13.84it/s] Update steps: 84%|████████████████████▏ | 8417/10000 [15:24<01:54, 13.85it/s] Update steps: 84%|████████████████████▏ | 8419/10000 [15:24<01:53, 13.91it/s] Update steps: 84%|████████████████████▏ | 8421/10000 [15:24<01:53, 13.88it/s] Update steps: 84%|████████████████████▏ | 8423/10000 [15:24<01:53, 13.90it/s] Update steps: 84%|████████████████████▏ | 8425/10000 [15:24<01:53, 13.91it/s] Update steps: 84%|████████████████████▏ | 8427/10000 [15:24<01:53, 13.91it/s] Update steps: 84%|████████████████████▏ | 8429/10000 [15:25<01:53, 13.87it/s] Update steps: 84%|████████████████████▏ | 8431/10000 [15:25<01:53, 13.81it/s] Update steps: 84%|████████████████████▏ | 8433/10000 [15:25<01:53, 13.82it/s] Update steps: 84%|████████████████████▏ | 8435/10000 [15:25<01:52, 13.88it/s] Update steps: 84%|████████████████████▏ | 8437/10000 [15:25<01:52, 13.86it/s] Update steps: 84%|████████████████████▎ | 8439/10000 [15:25<01:52, 13.91it/s] Update steps: 84%|████████████████████▎ | 8441/10000 [15:25<01:52, 13.89it/s] Update steps: 84%|████████████████████▎ | 8443/10000 [15:26<01:52, 13.87it/s] Update steps: 84%|████████████████████▎ | 8445/10000 [15:26<01:51, 13.94it/s] Update steps: 84%|████████████████████▎ | 8447/10000 [15:26<01:51, 13.99it/s] Update steps: 84%|████████████████████▎ | 8449/10000 [15:26<01:51, 13.95it/s] Update steps: 85%|████████████████████▎ | 8451/10000 [15:26<01:51, 13.89it/s] Update steps: 85%|████████████████████▎ | 8453/10000 [15:26<01:53, 13.60it/s] Update steps: 85%|████████████████████▎ | 8455/10000 [15:26<01:53, 13.67it/s] Update steps: 85%|████████████████████▎ | 8457/10000 [15:27<01:53, 13.57it/s] Update steps: 85%|████████████████████▎ | 8459/10000 [15:27<01:54, 13.51it/s] Update steps: 85%|████████████████████▎ | 8461/10000 [15:27<01:54, 13.46it/s] Update steps: 85%|████████████████████▎ | 8463/10000 [15:27<01:53, 13.50it/s] Update steps: 85%|████████████████████▎ | 8465/10000 [15:27<01:53, 13.49it/s] Update steps: 85%|████████████████████▎ | 8467/10000 [15:27<01:52, 13.63it/s] Update steps: 85%|████████████████████▎ | 8469/10000 [15:27<01:51, 13.69it/s] Update steps: 85%|████████████████████▎ | 8471/10000 [15:28<01:51, 13.77it/s] Update steps: 85%|████████████████████▎ | 8473/10000 [15:28<01:50, 13.84it/s] Update steps: 85%|████████████████████▎ | 8475/10000 [15:28<01:50, 13.86it/s] Update steps: 85%|████████████████████▎ | 8477/10000 [15:28<01:50, 13.78it/s] Update steps: 85%|████████████████████▎ | 8479/10000 [15:28<01:50, 13.76it/s] Update steps: 85%|████████████████████▎ | 8481/10000 [15:28<01:49, 13.82it/s] Update steps: 85%|████████████████████▎ | 8483/10000 [15:28<01:49, 13.82it/s] Update steps: 85%|████████████████████▎ | 8485/10000 [15:29<01:48, 13.93it/s] Update steps: 85%|████████████████████▎ | 8487/10000 [15:29<01:48, 13.95it/s] Update steps: 85%|████████████████████▎ | 8489/10000 [15:29<01:48, 13.99it/s] Update steps: 85%|████████████████████▍ | 8491/10000 [15:29<01:47, 13.99it/s] Update steps: 85%|████████████████████▍ | 8493/10000 [15:29<01:47, 14.04it/s] Update steps: 85%|████████████████████▍ | 8495/10000 [15:29<01:47, 14.03it/s] Update steps: 85%|████████████████████▍ | 8497/10000 [15:29<01:47, 13.97it/s] Update steps: 85%|████████████████████▍ | 8499/10000 [15:30<01:47, 13.99it/s] Update steps: 85%|████████████████████▍ | 8501/10000 [15:30<01:48, 13.78it/s] Update steps: 85%|████████████████████▍ | 8503/10000 [15:30<01:48, 13.78it/s] Update steps: 85%|████████████████████▍ | 8505/10000 [15:30<01:47, 13.88it/s] Update steps: 85%|████████████████████▍ | 8507/10000 [15:30<01:47, 13.85it/s] Update steps: 85%|████████████████████▍ | 8509/10000 [15:30<01:47, 13.88it/s] Update steps: 85%|████████████████████▍ | 8511/10000 [15:30<01:47, 13.91it/s] Update steps: 85%|████████████████████▍ | 8513/10000 [15:31<01:46, 13.91it/s] Update steps: 85%|████████████████████▍ | 8515/10000 [15:31<01:46, 13.96it/s] Update steps: 85%|████████████████████▍ | 8517/10000 [15:31<01:45, 13.99it/s] Update steps: 85%|████████████████████▍ | 8519/10000 [15:31<01:46, 13.86it/s] Update steps: 85%|████████████████████▍ | 8521/10000 [15:31<01:46, 13.89it/s] Update steps: 85%|████████████████████▍ | 8523/10000 [15:31<01:46, 13.90it/s] Update steps: 85%|████████████████████▍ | 8525/10000 [15:31<01:45, 13.93it/s] Update steps: 85%|████████████████████▍ | 8527/10000 [15:32<01:45, 13.93it/s] Update steps: 85%|████████████████████▍ | 8529/10000 [15:32<01:45, 13.94it/s] Update steps: 85%|████████████████████▍ | 8531/10000 [15:32<01:45, 13.93it/s] Update steps: 85%|████████████████████▍ | 8533/10000 [15:32<01:45, 13.90it/s] Update steps: 85%|████████████████████▍ | 8535/10000 [15:32<01:45, 13.82it/s] Update steps: 85%|████████████████████▍ | 8537/10000 [15:32<01:46, 13.78it/s] Update steps: 85%|████████████████████▍ | 8539/10000 [15:33<01:46, 13.74it/s] Update steps: 85%|████████████████████▍ | 8541/10000 [15:33<01:45, 13.82it/s] Update steps: 85%|████████████████████▌ | 8543/10000 [15:33<01:45, 13.87it/s] Update steps: 85%|████████████████████▌ | 8545/10000 [15:33<01:44, 13.92it/s] Update steps: 85%|████████████████████▌ | 8547/10000 [15:33<01:44, 13.91it/s] Update steps: 85%|████████████████████▌ | 8549/10000 [15:33<01:44, 13.95it/s] Update steps: 86%|████████████████████▌ | 8551/10000 [15:33<01:44, 13.87it/s] Update steps: 86%|████████████████████▌ | 8553/10000 [15:34<01:44, 13.88it/s] Update steps: 86%|████████████████████▌ | 8555/10000 [15:34<01:44, 13.85it/s] Update steps: 86%|████████████████████▌ | 8557/10000 [15:34<01:44, 13.86it/s] Update steps: 86%|████████████████████▌ | 8559/10000 [15:34<01:44, 13.77it/s] Update steps: 86%|████████████████████▌ | 8561/10000 [15:34<01:44, 13.82it/s] Update steps: 86%|████████████████████▌ | 8563/10000 [15:34<01:43, 13.88it/s] Update steps: 86%|████████████████████▌ | 8565/10000 [15:34<01:43, 13.85it/s] Update steps: 86%|████████████████████▌ | 8567/10000 [15:35<01:43, 13.91it/s] Update steps: 86%|████████████████████▌ | 8569/10000 [15:35<01:42, 13.89it/s] Update steps: 86%|████████████████████▌ | 8571/10000 [15:35<01:44, 13.71it/s] Update steps: 86%|████████████████████▌ | 8573/10000 [15:35<01:43, 13.80it/s] Update steps: 86%|████████████████████▌ | 8575/10000 [15:35<01:43, 13.83it/s] Update steps: 86%|████████████████████▌ | 8577/10000 [15:35<01:42, 13.94it/s] Update steps: 86%|████████████████████▌ | 8579/10000 [15:35<01:42, 13.90it/s] Update steps: 86%|████████████████████▌ | 8581/10000 [15:36<01:42, 13.84it/s] Update steps: 86%|████████████████████▌ | 8583/10000 [15:36<01:42, 13.88it/s] Update steps: 86%|████████████████████▌ | 8585/10000 [15:36<01:42, 13.82it/s] Update steps: 86%|████████████████████▌ | 8587/10000 [15:36<01:41, 13.92it/s] Update steps: 86%|████████████████████▌ | 8589/10000 [15:36<01:42, 13.83it/s] Update steps: 86%|████████████████████▌ | 8591/10000 [15:36<01:41, 13.84it/s] Update steps: 86%|████████████████████▌ | 8593/10000 [15:36<01:41, 13.89it/s] Update steps: 86%|████████████████████▋ | 8595/10000 [15:37<01:40, 13.96it/s] Update steps: 86%|████████████████████▋ | 8597/10000 [15:37<01:41, 13.79it/s] Update steps: 86%|████████████████████▋ | 8599/10000 [15:37<01:42, 13.70it/s] Update steps: 86%|████████████████████▋ | 8601/10000 [15:37<01:42, 13.63it/s] Update steps: 86%|████████████████████▋ | 8603/10000 [15:37<01:42, 13.60it/s] Update steps: 86%|████████████████████▋ | 8605/10000 [15:37<01:41, 13.73it/s] Update steps: 86%|████████████████████▋ | 8607/10000 [15:37<01:40, 13.80it/s] Update steps: 86%|████████████████████▋ | 8609/10000 [15:38<01:41, 13.73it/s] Update steps: 86%|████████████████████▋ | 8611/10000 [15:38<01:40, 13.80it/s] Update steps: 86%|████████████████████▋ | 8613/10000 [15:38<01:39, 13.88it/s] Update steps: 86%|████████████████████▋ | 8615/10000 [15:38<01:39, 13.97it/s] Update steps: 86%|████████████████████▋ | 8617/10000 [15:38<01:38, 13.98it/s] Update steps: 86%|████████████████████▋ | 8619/10000 [15:38<01:38, 13.99it/s] Update steps: 86%|████████████████████▋ | 8621/10000 [15:38<01:38, 14.02it/s] Update steps: 86%|████████████████████▋ | 8623/10000 [15:39<01:38, 14.05it/s] Update steps: 86%|████████████████████▋ | 8625/10000 [15:39<01:37, 14.04it/s] Update steps: 86%|████████████████████▋ | 8627/10000 [15:39<01:37, 14.06it/s] Update steps: 86%|████████████████████▋ | 8629/10000 [15:39<01:37, 14.11it/s] Update steps: 86%|████████████████████▋ | 8631/10000 [15:39<01:37, 14.10it/s] Update steps: 86%|████████████████████▋ | 8633/10000 [15:39<01:37, 14.07it/s] Update steps: 86%|████████████████████▋ | 8635/10000 [15:39<01:36, 14.09it/s] Update steps: 86%|████████████████████▋ | 8637/10000 [15:40<01:37, 13.94it/s] Update steps: 86%|████████████████████▋ | 8639/10000 [15:40<01:37, 13.96it/s] Update steps: 86%|████████████████████▋ | 8641/10000 [15:40<01:37, 13.93it/s] Update steps: 86%|████████████████████▋ | 8643/10000 [15:40<01:38, 13.81it/s] Update steps: 86%|████████████████████▋ | 8645/10000 [15:40<01:38, 13.77it/s] Update steps: 86%|████████████████████▊ | 8647/10000 [15:40<01:38, 13.71it/s] Update steps: 86%|████████████████████▊ | 8649/10000 [15:40<01:37, 13.81it/s] Update steps: 87%|████████████████████▊ | 8651/10000 [15:41<01:37, 13.87it/s] Update steps: 87%|████████████████████▊ | 8653/10000 [15:41<01:37, 13.84it/s] Update steps: 87%|████████████████████▊ | 8655/10000 [15:41<01:37, 13.85it/s] Update steps: 87%|████████████████████▊ | 8657/10000 [15:41<01:37, 13.83it/s] Update steps: 87%|████████████████████▊ | 8659/10000 [15:41<01:36, 13.86it/s] Update steps: 87%|████████████████████▊ | 8661/10000 [15:41<01:36, 13.90it/s] Update steps: 87%|████████████████████▊ | 8663/10000 [15:41<01:35, 13.96it/s] Update steps: 87%|████████████████████▊ | 8665/10000 [15:42<01:36, 13.90it/s] Update steps: 87%|████████████████████▊ | 8667/10000 [15:42<01:35, 13.96it/s] Update steps: 87%|████████████████████▊ | 8669/10000 [15:42<01:35, 13.94it/s] Update steps: 87%|████████████████████▊ | 8671/10000 [15:42<01:35, 13.93it/s] Update steps: 87%|████████████████████▊ | 8673/10000 [15:42<01:35, 13.95it/s] Update steps: 87%|████████████████████▊ | 8675/10000 [15:42<01:35, 13.87it/s] Update steps: 87%|████████████████████▊ | 8677/10000 [15:42<01:35, 13.86it/s] Update steps: 87%|████████████████████▊ | 8679/10000 [15:43<01:35, 13.78it/s] Update steps: 87%|████████████████████▊ | 8681/10000 [15:43<01:37, 13.55it/s] Update steps: 87%|████████████████████▊ | 8683/10000 [15:43<01:36, 13.67it/s] Update steps: 87%|████████████████████▊ | 8685/10000 [15:43<01:35, 13.72it/s] Update steps: 87%|████████████████████▊ | 8687/10000 [15:43<01:35, 13.82it/s] Update steps: 87%|████████████████████▊ | 8689/10000 [15:43<01:34, 13.87it/s] Update steps: 87%|████████████████████▊ | 8691/10000 [15:43<01:33, 13.94it/s] Update steps: 87%|████████████████████▊ | 8693/10000 [15:44<01:33, 13.97it/s] Update steps: 87%|████████████████████▊ | 8695/10000 [15:44<01:33, 13.98it/s] Update steps: 87%|████████████████████▊ | 8697/10000 [15:44<01:33, 13.87it/s] Update steps: 87%|████████████████████▉ | 8699/10000 [15:44<01:33, 13.91it/s] Update steps: 87%|████████████████████▉ | 8701/10000 [15:44<01:33, 13.92it/s] Update steps: 87%|████████████████████▉ | 8703/10000 [15:44<01:34, 13.70it/s] Update steps: 87%|████████████████████▉ | 8705/10000 [15:44<01:34, 13.70it/s] Update steps: 87%|████████████████████▉ | 8707/10000 [15:45<01:34, 13.73it/s] Update steps: 87%|████████████████████▉ | 8709/10000 [15:45<01:35, 13.53it/s] Update steps: 87%|████████████████████▉ | 8711/10000 [15:45<01:34, 13.64it/s] Update steps: 87%|████████████████████▉ | 8713/10000 [15:45<01:33, 13.75it/s] Update steps: 87%|████████████████████▉ | 8715/10000 [15:45<01:33, 13.76it/s] Update steps: 87%|████████████████████▉ | 8717/10000 [15:45<01:35, 13.50it/s] Update steps: 87%|████████████████████▉ | 8719/10000 [15:46<01:34, 13.53it/s] Update steps: 87%|████████████████████▉ | 8721/10000 [15:46<01:34, 13.60it/s] Update steps: 87%|████████████████████▉ | 8723/10000 [15:46<01:33, 13.67it/s] Update steps: 87%|████████████████████▉ | 8725/10000 [15:46<01:32, 13.82it/s] Update steps: 87%|████████████████████▉ | 8727/10000 [15:46<01:31, 13.90it/s] Update steps: 87%|████████████████████▉ | 8729/10000 [15:46<01:31, 13.92it/s] Update steps: 87%|████████████████████▉ | 8731/10000 [15:46<01:30, 13.95it/s] Update steps: 87%|████████████████████▉ | 8733/10000 [15:47<01:30, 13.99it/s] Update steps: 87%|████████████████████▉ | 8735/10000 [15:47<01:30, 13.98it/s] Update steps: 87%|████████████████████▉ | 8737/10000 [15:47<01:30, 14.01it/s] Update steps: 87%|████████████████████▉ | 8739/10000 [15:47<01:30, 13.96it/s] Update steps: 87%|████████████████████▉ | 8741/10000 [15:47<01:29, 14.00it/s] Update steps: 87%|████████████████████▉ | 8743/10000 [15:47<01:30, 13.88it/s] Update steps: 87%|████████████████████▉ | 8745/10000 [15:47<01:30, 13.94it/s] Update steps: 87%|████████████████████▉ | 8747/10000 [15:48<01:29, 13.95it/s] Update steps: 87%|████████████████████▉ | 8749/10000 [15:48<01:29, 13.93it/s] Update steps: 88%|█████████████████████ | 8751/10000 [15:48<01:30, 13.83it/s] Update steps: 88%|█████████████████████ | 8753/10000 [15:48<01:30, 13.80it/s] Update steps: 88%|█████████████████████ | 8755/10000 [15:48<01:29, 13.87it/s] Update steps: 88%|█████████████████████ | 8757/10000 [15:48<01:29, 13.95it/s] Update steps: 88%|█████████████████████ | 8759/10000 [15:48<01:28, 13.97it/s] Update steps: 88%|█████████████████████ | 8761/10000 [15:49<01:29, 13.80it/s] Update steps: 88%|█████████████████████ | 8763/10000 [15:49<01:30, 13.72it/s] Update steps: 88%|█████████████████████ | 8765/10000 [15:49<01:29, 13.83it/s] Update steps: 88%|█████████████████████ | 8767/10000 [15:49<01:28, 13.86it/s] Update steps: 88%|█████████████████████ | 8769/10000 [15:49<01:28, 13.93it/s] Update steps: 88%|█████████████████████ | 8771/10000 [15:49<01:29, 13.70it/s] Update steps: 88%|█████████████████████ | 8773/10000 [15:49<01:28, 13.80it/s] Update steps: 88%|█████████████████████ | 8775/10000 [15:50<01:28, 13.92it/s] Update steps: 88%|█████████████████████ | 8777/10000 [15:50<01:28, 13.89it/s] Update steps: 88%|█████████████████████ | 8779/10000 [15:50<01:27, 13.95it/s] Update steps: 88%|█████████████████████ | 8781/10000 [15:50<01:27, 13.97it/s] Update steps: 88%|█████████████████████ | 8783/10000 [15:50<01:26, 14.02it/s] Update steps: 88%|█████████████████████ | 8785/10000 [15:50<01:26, 14.01it/s] Update steps: 88%|█████████████████████ | 8787/10000 [15:50<01:27, 13.84it/s] Update steps: 88%|█████████████████████ | 8789/10000 [15:51<01:27, 13.91it/s] Update steps: 88%|█████████████████████ | 8791/10000 [15:51<01:27, 13.84it/s] Update steps: 88%|█████████████████████ | 8793/10000 [15:51<01:28, 13.70it/s] Update steps: 88%|█████████████████████ | 8795/10000 [15:51<01:27, 13.83it/s] Update steps: 88%|█████████████████████ | 8797/10000 [15:51<01:26, 13.88it/s] Update steps: 88%|█████████████████████ | 8799/10000 [15:51<01:26, 13.94it/s] Update steps: 88%|█████████████████████ | 8801/10000 [15:51<01:25, 13.98it/s] Update steps: 88%|█████████████████████▏ | 8803/10000 [15:52<01:26, 13.90it/s] Update steps: 88%|█████████████████████▏ | 8805/10000 [15:52<01:26, 13.89it/s] Update steps: 88%|█████████████████████▏ | 8807/10000 [15:52<01:25, 13.93it/s] Update steps: 88%|█████████████████████▏ | 8809/10000 [15:52<01:25, 13.96it/s] Update steps: 88%|█████████████████████▏ | 8811/10000 [15:52<01:25, 13.95it/s] Update steps: 88%|█████████████████████▏ | 8813/10000 [15:52<01:25, 13.93it/s] Update steps: 88%|█████████████████████▏ | 8815/10000 [15:52<01:26, 13.78it/s] Update steps: 88%|█████████████████████▏ | 8817/10000 [15:53<01:25, 13.88it/s] Update steps: 88%|█████████████████████▏ | 8819/10000 [15:53<01:25, 13.84it/s] Update steps: 88%|█████████████████████▏ | 8821/10000 [15:53<01:25, 13.74it/s] Update steps: 88%|█████████████████████▏ | 8823/10000 [15:53<01:25, 13.78it/s] Update steps: 88%|█████████████████████▏ | 8825/10000 [15:53<01:25, 13.69it/s] Update steps: 88%|█████████████████████▏ | 8827/10000 [15:53<01:25, 13.65it/s] Update steps: 88%|█████████████████████▏ | 8829/10000 [15:53<01:25, 13.77it/s] Update steps: 88%|█████████████████████▏ | 8831/10000 [15:54<01:24, 13.90it/s] Update steps: 88%|█████████████████████▏ | 8833/10000 [15:54<01:23, 13.96it/s] Update steps: 88%|█████████████████████▏ | 8835/10000 [15:54<01:23, 13.89it/s] Update steps: 88%|█████████████████████▏ | 8837/10000 [15:54<01:23, 13.94it/s] Update steps: 88%|█████████████████████▏ | 8839/10000 [15:54<01:23, 13.94it/s] Update steps: 88%|█████████████████████▏ | 8841/10000 [15:54<01:23, 13.95it/s] Update steps: 88%|█████████████████████▏ | 8843/10000 [15:54<01:23, 13.94it/s] Update steps: 88%|█████████████████████▏ | 8845/10000 [15:55<01:22, 13.96it/s] Update steps: 88%|█████████████████████▏ | 8847/10000 [15:55<01:23, 13.76it/s] Update steps: 88%|█████████████████████▏ | 8849/10000 [15:55<01:23, 13.80it/s] Update steps: 89%|█████████████████████▏ | 8851/10000 [15:55<01:23, 13.81it/s] Update steps: 89%|█████████████████████▏ | 8853/10000 [15:55<01:22, 13.82it/s] Update steps: 89%|█████████████████████▎ | 8855/10000 [15:55<01:22, 13.87it/s] Update steps: 89%|█████████████████████▎ | 8857/10000 [15:55<01:22, 13.93it/s] Update steps: 89%|█████████████████████▎ | 8859/10000 [15:56<01:21, 13.97it/s] Update steps: 89%|█████████████████████▎ | 8861/10000 [15:56<01:21, 13.91it/s] Update steps: 89%|█████████████████████▎ | 8863/10000 [15:56<01:21, 13.93it/s] Update steps: 89%|█████████████████████▎ | 8865/10000 [15:56<01:24, 13.50it/s] Update steps: 89%|█████████████████████▎ | 8867/10000 [15:56<01:22, 13.68it/s] Update steps: 89%|█████████████████████▎ | 8869/10000 [15:56<01:22, 13.78it/s] Update steps: 89%|█████████████████████▎ | 8871/10000 [15:56<01:21, 13.79it/s] Update steps: 89%|█████████████████████▎ | 8873/10000 [15:57<01:21, 13.89it/s] Update steps: 89%|█████████████████████▎ | 8875/10000 [15:57<01:21, 13.87it/s] Update steps: 89%|█████████████████████▎ | 8877/10000 [15:57<01:20, 13.91it/s] Update steps: 89%|█████████████████████▎ | 8879/10000 [15:57<01:21, 13.82it/s] Update steps: 89%|█████████████████████▎ | 8881/10000 [15:57<01:20, 13.88it/s] Update steps: 89%|█████████████████████▎ | 8883/10000 [15:57<01:20, 13.85it/s] Update steps: 89%|█████████████████████▎ | 8885/10000 [15:57<01:20, 13.91it/s] Update steps: 89%|█████████████████████▎ | 8887/10000 [15:58<01:20, 13.86it/s] Update steps: 89%|█████████████████████▎ | 8889/10000 [15:58<01:19, 13.91it/s] Update steps: 89%|█████████████████████▎ | 8891/10000 [15:58<01:19, 13.91it/s] Update steps: 89%|█████████████████████▎ | 8893/10000 [15:58<01:19, 13.97it/s] Update steps: 89%|█████████████████████▎ | 8895/10000 [15:58<01:19, 13.92it/s] Update steps: 89%|█████████████████████▎ | 8897/10000 [15:58<01:20, 13.63it/s] Update steps: 89%|█████████████████████▎ | 8899/10000 [15:58<01:21, 13.51it/s] Update steps: 89%|█████████████████████▎ | 8901/10000 [15:59<01:21, 13.56it/s] Update steps: 89%|█████████████████████▎ | 8903/10000 [15:59<01:20, 13.65it/s] Update steps: 89%|█████████████████████▎ | 8905/10000 [15:59<01:19, 13.72it/s] Update steps: 89%|█████████████████████▍ | 8907/10000 [15:59<01:19, 13.80it/s] Update steps: 89%|█████████████████████▍ | 8909/10000 [15:59<01:18, 13.91it/s] Update steps: 89%|█████████████████████▍ | 8911/10000 [15:59<01:18, 13.91it/s] Update steps: 89%|█████████████████████▍ | 8913/10000 [16:00<01:18, 13.79it/s] Update steps: 89%|█████████████████████▍ | 8915/10000 [16:00<01:19, 13.73it/s] Update steps: 89%|█████████████████████▍ | 8917/10000 [16:00<01:18, 13.81it/s] Update steps: 89%|█████████████████████▍ | 8919/10000 [16:00<01:18, 13.79it/s] Update steps: 89%|█████████████████████▍ | 8921/10000 [16:00<01:18, 13.82it/s] Update steps: 89%|█████████████████████▍ | 8923/10000 [16:00<01:17, 13.90it/s] Update steps: 89%|█████████████████████▍ | 8925/10000 [16:00<01:17, 13.88it/s] Update steps: 89%|█████████████████████▍ | 8927/10000 [16:01<01:17, 13.93it/s] Update steps: 89%|█████████████████████▍ | 8929/10000 [16:01<01:17, 13.84it/s] Update steps: 89%|█████████████████████▍ | 8931/10000 [16:01<01:17, 13.83it/s] Update steps: 89%|█████████████████████▍ | 8933/10000 [16:01<01:17, 13.81it/s] Update steps: 89%|█████████████████████▍ | 8935/10000 [16:01<01:16, 13.85it/s] Update steps: 89%|█████████████████████▍ | 8937/10000 [16:01<01:16, 13.85it/s] Update steps: 89%|█████████████████████▍ | 8939/10000 [16:01<01:16, 13.93it/s] Update steps: 89%|█████████████████████▍ | 8941/10000 [16:02<01:15, 13.97it/s] Update steps: 89%|█████████████████████▍ | 8943/10000 [16:02<01:15, 13.91it/s] Update steps: 89%|█████████████████████▍ | 8945/10000 [16:02<01:15, 13.95it/s] Update steps: 89%|█████████████████████▍ | 8947/10000 [16:02<01:16, 13.83it/s] Update steps: 89%|█████████████████████▍ | 8949/10000 [16:02<01:15, 13.90it/s] Update steps: 90%|█████████████████████▍ | 8951/10000 [16:02<01:15, 13.91it/s] Update steps: 90%|█████████████████████▍ | 8953/10000 [16:02<01:14, 13.96it/s] Update steps: 90%|█████████████████████▍ | 8955/10000 [16:03<01:14, 13.99it/s] Update steps: 90%|█████████████████████▍ | 8957/10000 [16:03<01:14, 13.99it/s] Update steps: 90%|█████████████████████▌ | 8959/10000 [16:03<01:14, 13.92it/s] Update steps: 90%|█████████████████████▌ | 8961/10000 [16:03<01:14, 13.91it/s] Update steps: 90%|█████████████████████▌ | 8963/10000 [16:03<01:14, 13.90it/s] Update steps: 90%|█████████████████████▌ | 8965/10000 [16:03<01:14, 13.90it/s] Update steps: 90%|█████████████████████▌ | 8967/10000 [16:03<01:14, 13.82it/s] Update steps: 90%|█████████████████████▌ | 8969/10000 [16:04<01:14, 13.89it/s] Update steps: 90%|█████████████████████▌ | 8971/10000 [16:04<01:14, 13.90it/s] Update steps: 90%|█████████████████████▌ | 8973/10000 [16:04<01:14, 13.84it/s] Update steps: 90%|█████████████████████▌ | 8975/10000 [16:04<01:13, 13.97it/s] Update steps: 90%|█████████████████████▌ | 8977/10000 [16:04<01:13, 13.92it/s] Update steps: 90%|█████████████████████▌ | 8979/10000 [16:04<01:12, 13.99it/s] Update steps: 90%|█████████████████████▌ | 8981/10000 [16:04<01:13, 13.96it/s] Update steps: 90%|█████████████████████▌ | 8983/10000 [16:05<01:13, 13.83it/s] Update steps: 90%|█████████████████████▌ | 8985/10000 [16:05<01:13, 13.76it/s] Update steps: 90%|█████████████████████▌ | 8987/10000 [16:05<01:13, 13.73it/s] Update steps: 90%|█████████████████████▌ | 8989/10000 [16:05<01:13, 13.78it/s] Update steps: 90%|█████████████████████▌ | 8991/10000 [16:05<01:13, 13.81it/s] Update steps: 90%|█████████████████████▌ | 8993/10000 [16:05<01:12, 13.88it/s] Update steps: 90%|█████████████████████▌ | 8995/10000 [16:05<01:12, 13.91it/s] Update steps: 90%|█████████████████████▌ | 8997/10000 [16:06<01:12, 13.93it/s] Update steps: 90%|█████████████████████▌ | 8999/10000 [16:06<01:11, 13.95it/s]2026-04-17 11:09:26.657 | INFO | __main__:main:490 - Performing evaluation at step 9000 +2026-04-17 11:09:26.659 | INFO | utils.eval:evaluate_model:21 - Loading validation data from 8 local validation files +2026-04-17 11:09:27.448 | INFO | utils.eval:evaluate_model:43 - Loaded validation dataset in 0.79 seconds +2026-04-17 11:09:27.449 | INFO | utils.eval:evaluate_model:60 - Eval set prepared in 0.79 seconds + Update steps: 90%|█████████████████████▌ | 9000/10000 [16:16<01:11, 13.95it/s]2026-04-17 11:10:03.331 | INFO | __main__:main:505 - Eval loss at step 9000: 3.4098527671716905, Eval perplexity: 30.260788549909698 + Update steps: 90%|███████████████████▊ | 9001/10000 [16:43<1:32:46, 5.57s/it] Update steps: 90%|███████████████████▊ | 9003/10000 [16:43<1:05:10, 3.92s/it] Update steps: 90%|█████████████████████▌ | 9005/10000 [16:43<45:52, 2.77s/it] Update steps: 90%|█████████████████████▌ | 9007/10000 [16:43<32:24, 1.96s/it] Update steps: 90%|█████████████████████▌ | 9009/10000 [16:43<22:59, 1.39s/it] Update steps: 90%|█████████████████████▋ | 9011/10000 [16:43<16:25, 1.00it/s] Update steps: 90%|█████████████████████▋ | 9013/10000 [16:43<11:50, 1.39it/s] Update steps: 90%|█████████████████████▋ | 9015/10000 [16:44<08:38, 1.90it/s] Update steps: 90%|█████████████████████▋ | 9017/10000 [16:44<06:22, 2.57it/s] Update steps: 90%|█████████████████████▋ | 9019/10000 [16:44<04:48, 3.40it/s] Update steps: 90%|█████████████████████▋ | 9021/10000 [16:44<03:42, 4.40it/s] Update steps: 90%|█████████████████████▋ | 9023/10000 [16:44<02:56, 5.52it/s] Update steps: 90%|█████████████████████▋ | 9025/10000 [16:44<02:24, 6.74it/s] Update steps: 90%|█████████████████████▋ | 9027/10000 [16:44<02:02, 7.96it/s] Update steps: 90%|█████████████████████▋ | 9029/10000 [16:45<01:46, 9.15it/s] Update steps: 90%|█████████████████████▋ | 9031/10000 [16:45<01:35, 10.15it/s] Update steps: 90%|█████████████████████▋ | 9033/10000 [16:45<01:27, 11.00it/s] Update steps: 90%|█████████████████████▋ | 9035/10000 [16:45<01:22, 11.71it/s] Update steps: 90%|█████████████████████▋ | 9037/10000 [16:45<01:18, 12.29it/s] Update steps: 90%|█████████████████████▋ | 9039/10000 [16:45<01:15, 12.76it/s] Update steps: 90%|█████████████████████▋ | 9041/10000 [16:45<01:13, 13.10it/s] Update steps: 90%|█████████████████████▋ | 9043/10000 [16:46<01:11, 13.36it/s] Update steps: 90%|█████████████████████▋ | 9045/10000 [16:46<01:10, 13.52it/s] Update steps: 90%|█████████████████████▋ | 9047/10000 [16:46<01:10, 13.57it/s] Update steps: 90%|█████████████████████▋ | 9049/10000 [16:46<01:10, 13.54it/s] Update steps: 91%|█████████████████████▋ | 9051/10000 [16:46<01:09, 13.64it/s] Update steps: 91%|█████████████████████▋ | 9053/10000 [16:46<01:08, 13.78it/s] Update steps: 91%|█████████████████████▋ | 9055/10000 [16:46<01:10, 13.49it/s] Update steps: 91%|█████████████████████▋ | 9057/10000 [16:47<01:09, 13.56it/s] Update steps: 91%|█████████████████████▋ | 9059/10000 [16:47<01:08, 13.70it/s] Update steps: 91%|█████████████████████▋ | 9061/10000 [16:47<01:08, 13.71it/s] Update steps: 91%|█████████████████████▊ | 9063/10000 [16:47<01:08, 13.68it/s] Update steps: 91%|█████████████████████▊ | 9065/10000 [16:47<01:07, 13.76it/s] Update steps: 91%|█████████████████████▊ | 9067/10000 [16:47<01:08, 13.72it/s] Update steps: 91%|█████████████████████▊ | 9069/10000 [16:47<01:07, 13.75it/s] Update steps: 91%|█████████████████████▊ | 9071/10000 [16:48<01:07, 13.79it/s] Update steps: 91%|█████████████████████▊ | 9073/10000 [16:48<01:06, 13.86it/s] Update steps: 91%|█████████████████████▊ | 9075/10000 [16:48<01:07, 13.75it/s] Update steps: 91%|█████████████████████▊ | 9077/10000 [16:48<01:07, 13.72it/s] Update steps: 91%|█████████████████████▊ | 9079/10000 [16:48<01:07, 13.69it/s] Update steps: 91%|█████████████████████▊ | 9081/10000 [16:48<01:07, 13.63it/s] Update steps: 91%|█████████████████████▊ | 9083/10000 [16:48<01:06, 13.77it/s] Update steps: 91%|█████████████████████▊ | 9085/10000 [16:49<01:06, 13.80it/s] Update steps: 91%|█████████████████████▊ | 9087/10000 [16:49<01:05, 13.86it/s] Update steps: 91%|█████████████████████▊ | 9089/10000 [16:49<01:06, 13.79it/s] Update steps: 91%|█████████████████████▊ | 9091/10000 [16:49<01:05, 13.85it/s] Update steps: 91%|█████████████████████▊ | 9093/10000 [16:49<01:05, 13.89it/s] Update steps: 91%|█████████████████████▊ | 9095/10000 [16:49<01:06, 13.68it/s] Update steps: 91%|█████████████████████▊ | 9097/10000 [16:49<01:05, 13.81it/s] Update steps: 91%|█████████████████████▊ | 9099/10000 [16:50<01:05, 13.84it/s] Update steps: 91%|█████████████████████▊ | 9101/10000 [16:50<01:04, 13.91it/s] Update steps: 91%|█████████████████████▊ | 9103/10000 [16:50<01:04, 13.95it/s] Update steps: 91%|█████████████████████▊ | 9105/10000 [16:50<01:04, 13.92it/s] Update steps: 91%|█████████████████████▊ | 9107/10000 [16:50<01:04, 13.81it/s] Update steps: 91%|█████████████████████▊ | 9109/10000 [16:50<01:04, 13.73it/s] Update steps: 91%|█████████████████████▊ | 9111/10000 [16:50<01:04, 13.77it/s] Update steps: 91%|█████████████████████▊ | 9113/10000 [16:51<01:04, 13.83it/s] Update steps: 91%|█████████████████████▉ | 9115/10000 [16:51<01:03, 13.93it/s] Update steps: 91%|█████████████████████▉ | 9117/10000 [16:51<01:03, 13.82it/s] Update steps: 91%|█████████████████████▉ | 9119/10000 [16:51<01:03, 13.81it/s] Update steps: 91%|█████████████████████▉ | 9121/10000 [16:51<01:04, 13.70it/s] Update steps: 91%|█████████████████████▉ | 9123/10000 [16:51<01:04, 13.70it/s] Update steps: 91%|█████████████████████▉ | 9125/10000 [16:51<01:03, 13.73it/s] Update steps: 91%|█████████████████████▉ | 9127/10000 [16:52<01:03, 13.81it/s] Update steps: 91%|█████████████████████▉ | 9129/10000 [16:52<01:03, 13.76it/s] Update steps: 91%|█████████████████████▉ | 9131/10000 [16:52<01:03, 13.76it/s] Update steps: 91%|█████████████████████▉ | 9133/10000 [16:52<01:02, 13.81it/s] Update steps: 91%|█████████████████████▉ | 9135/10000 [16:52<01:02, 13.88it/s] Update steps: 91%|█████████████████████▉ | 9137/10000 [16:52<01:04, 13.46it/s] Update steps: 91%|█████████████████████▉ | 9139/10000 [16:53<01:03, 13.64it/s] Update steps: 91%|█████████████████████▉ | 9141/10000 [16:53<01:03, 13.52it/s] Update steps: 91%|█████████████████████▉ | 9143/10000 [16:53<01:02, 13.66it/s] Update steps: 91%|█████████████████████▉ | 9145/10000 [16:53<01:02, 13.77it/s] Update steps: 91%|█████████████████████▉ | 9147/10000 [16:53<01:02, 13.75it/s] Update steps: 91%|█████████████████████▉ | 9149/10000 [16:53<01:02, 13.67it/s] Update steps: 92%|█████████████████████▉ | 9151/10000 [16:53<01:01, 13.72it/s] Update steps: 92%|█████████████████████▉ | 9153/10000 [16:54<01:01, 13.78it/s] Update steps: 92%|█████████████████████▉ | 9155/10000 [16:54<01:01, 13.80it/s] Update steps: 92%|█████████████████████▉ | 9157/10000 [16:54<01:00, 13.88it/s] Update steps: 92%|█████████████████████▉ | 9159/10000 [16:54<01:00, 13.94it/s] Update steps: 92%|█████████████████████▉ | 9161/10000 [16:54<01:01, 13.71it/s] Update steps: 92%|█████████████████████▉ | 9163/10000 [16:54<01:01, 13.71it/s] Update steps: 92%|█████████████████████▉ | 9165/10000 [16:54<01:03, 13.17it/s] Update steps: 92%|██████████████████████ | 9167/10000 [16:55<01:02, 13.39it/s] Update steps: 92%|██████████████████████ | 9169/10000 [16:55<01:01, 13.44it/s] Update steps: 92%|██████████████████████ | 9171/10000 [16:55<01:00, 13.62it/s] Update steps: 92%|██████████████████████ | 9173/10000 [16:55<01:00, 13.74it/s] Update steps: 92%|██████████████████████ | 9175/10000 [16:55<00:59, 13.79it/s] Update steps: 92%|██████████████████████ | 9177/10000 [16:55<00:59, 13.84it/s] Update steps: 92%|██████████████████████ | 9179/10000 [16:55<00:59, 13.78it/s] Update steps: 92%|██████████████████████ | 9181/10000 [16:56<00:59, 13.86it/s] Update steps: 92%|██████████████████████ | 9183/10000 [16:56<00:59, 13.77it/s] Update steps: 92%|██████████████████████ | 9185/10000 [16:56<00:59, 13.81it/s] Update steps: 92%|██████████████████████ | 9187/10000 [16:56<00:59, 13.73it/s] Update steps: 92%|██████████████████████ | 9189/10000 [16:56<00:58, 13.82it/s] Update steps: 92%|██████████████████████ | 9191/10000 [16:56<00:58, 13.84it/s] Update steps: 92%|██████████████████████ | 9193/10000 [16:56<00:57, 13.92it/s] Update steps: 92%|██████████████████████ | 9195/10000 [16:57<00:58, 13.82it/s] Update steps: 92%|██████████████████████ | 9197/10000 [16:57<00:58, 13.81it/s] Update steps: 92%|██████████████████████ | 9199/10000 [16:57<00:57, 13.83it/s] Update steps: 92%|██████████████████████ | 9201/10000 [16:57<00:57, 13.89it/s] Update steps: 92%|██████████████████████ | 9203/10000 [16:57<00:58, 13.67it/s] Update steps: 92%|██████████████████████ | 9205/10000 [16:57<00:58, 13.67it/s] Update steps: 92%|██████████████████████ | 9207/10000 [16:57<00:58, 13.60it/s] Update steps: 92%|██████████████████████ | 9209/10000 [16:58<00:58, 13.53it/s] Update steps: 92%|██████████████████████ | 9211/10000 [16:58<00:58, 13.58it/s] Update steps: 92%|██████████████████████ | 9213/10000 [16:58<00:57, 13.60it/s] Update steps: 92%|██████████████████████ | 9215/10000 [16:58<00:57, 13.67it/s] Update steps: 92%|██████████████████████ | 9217/10000 [16:58<00:56, 13.76it/s] Update steps: 92%|██████████████████████▏ | 9219/10000 [16:58<00:56, 13.83it/s] Update steps: 92%|██████████████████████▏ | 9221/10000 [16:58<00:56, 13.76it/s] Update steps: 92%|██████████████████████▏ | 9223/10000 [16:59<00:56, 13.64it/s] Update steps: 92%|██████████████████████▏ | 9225/10000 [16:59<00:56, 13.70it/s] Update steps: 92%|██████████████████████▏ | 9227/10000 [16:59<00:56, 13.80it/s] Update steps: 92%|██████████████████████▏ | 9229/10000 [16:59<00:55, 13.87it/s] Update steps: 92%|██████████████████████▏ | 9231/10000 [16:59<00:56, 13.63it/s] Update steps: 92%|██████████████████████▏ | 9233/10000 [16:59<00:56, 13.68it/s] Update steps: 92%|██████████████████████▏ | 9235/10000 [17:00<00:55, 13.78it/s] Update steps: 92%|██████████████████████▏ | 9237/10000 [17:00<00:55, 13.82it/s] Update steps: 92%|██████████████████████▏ | 9239/10000 [17:00<00:54, 13.88it/s] Update steps: 92%|██████████████████████▏ | 9241/10000 [17:00<00:54, 13.90it/s] Update steps: 92%|██████████████████████▏ | 9243/10000 [17:00<00:54, 13.80it/s] Update steps: 92%|██████████████████████▏ | 9245/10000 [17:00<00:54, 13.82it/s] Update steps: 92%|██████████████████████▏ | 9247/10000 [17:00<00:54, 13.89it/s] Update steps: 92%|██████████████████████▏ | 9249/10000 [17:01<00:53, 13.93it/s] Update steps: 93%|██████████████████████▏ | 9251/10000 [17:01<00:53, 13.99it/s] Update steps: 93%|██████████████████████▏ | 9253/10000 [17:01<00:53, 13.96it/s] Update steps: 93%|██████████████████████▏ | 9255/10000 [17:01<00:53, 13.92it/s] Update steps: 93%|██████████████████████▏ | 9257/10000 [17:01<00:53, 13.92it/s] Update steps: 93%|██████████████████████▏ | 9259/10000 [17:01<00:53, 13.88it/s] Update steps: 93%|██████████████████████▏ | 9261/10000 [17:01<00:52, 13.95it/s] Update steps: 93%|██████████████████████▏ | 9263/10000 [17:02<00:52, 14.01it/s] Update steps: 93%|██████████████████████▏ | 9265/10000 [17:02<00:52, 13.96it/s] Update steps: 93%|██████████████████████▏ | 9267/10000 [17:02<00:52, 13.99it/s] Update steps: 93%|██████████████████████▏ | 9269/10000 [17:02<00:52, 13.94it/s] Update steps: 93%|██████████████████████▎ | 9271/10000 [17:02<00:52, 13.92it/s] Update steps: 93%|██████████████████████▎ | 9273/10000 [17:02<00:52, 13.96it/s] Update steps: 93%|██████████████████████▎ | 9275/10000 [17:02<00:51, 13.98it/s] Update steps: 93%|██████████████████████▎ | 9277/10000 [17:03<00:51, 13.99it/s] Update steps: 93%|██████████████████████▎ | 9279/10000 [17:03<00:51, 13.91it/s] Update steps: 93%|██████████████████████▎ | 9281/10000 [17:03<00:51, 13.84it/s] Update steps: 93%|██████████████████████▎ | 9283/10000 [17:03<00:51, 13.81it/s] Update steps: 93%|██████████████████████▎ | 9285/10000 [17:03<00:52, 13.75it/s] Update steps: 93%|██████████████████████▎ | 9287/10000 [17:03<00:51, 13.84it/s] Update steps: 93%|██████████████████████▎ | 9289/10000 [17:03<00:51, 13.91it/s] Update steps: 93%|██████████████████████▎ | 9291/10000 [17:04<00:51, 13.87it/s] Update steps: 93%|██████████████████████▎ | 9293/10000 [17:04<00:50, 13.95it/s] Update steps: 93%|██████████████████████▎ | 9295/10000 [17:04<00:50, 13.92it/s] Update steps: 93%|██████████████████████▎ | 9297/10000 [17:04<00:50, 13.91it/s] Update steps: 93%|██████████████████████▎ | 9299/10000 [17:04<00:50, 13.98it/s] Update steps: 93%|██████████████████████▎ | 9301/10000 [17:04<00:50, 13.91it/s] Update steps: 93%|██████████████████████▎ | 9303/10000 [17:04<00:49, 13.96it/s] Update steps: 93%|██████████████████████▎ | 9305/10000 [17:05<00:49, 13.93it/s] Update steps: 93%|██████████████████████▎ | 9307/10000 [17:05<00:49, 13.89it/s] Update steps: 93%|██████████████████████▎ | 9309/10000 [17:05<00:50, 13.77it/s] Update steps: 93%|██████████████████████▎ | 9311/10000 [17:05<00:49, 13.80it/s] Update steps: 93%|██████████████████████▎ | 9313/10000 [17:05<00:49, 13.83it/s] Update steps: 93%|██████████████████████▎ | 9315/10000 [17:05<00:49, 13.78it/s] Update steps: 93%|██████████████████████▎ | 9317/10000 [17:05<00:49, 13.77it/s] Update steps: 93%|██████████████████████▎ | 9319/10000 [17:06<00:49, 13.86it/s] Update steps: 93%|██████████████████████▎ | 9321/10000 [17:06<00:48, 13.93it/s] Update steps: 93%|██████████████████████▍ | 9323/10000 [17:06<00:48, 13.88it/s] Update steps: 93%|██████████████████████▍ | 9325/10000 [17:06<00:48, 13.83it/s] Update steps: 93%|██████████████████████▍ | 9327/10000 [17:06<00:48, 13.79it/s] Update steps: 93%|██████████████████████▍ | 9329/10000 [17:06<00:48, 13.88it/s] Update steps: 93%|██████████████████████▍ | 9331/10000 [17:06<00:48, 13.84it/s] Update steps: 93%|██████████████████████▍ | 9333/10000 [17:07<00:48, 13.87it/s] Update steps: 93%|██████████████████████▍ | 9335/10000 [17:07<00:47, 13.87it/s] Update steps: 93%|██████████████████████▍ | 9337/10000 [17:07<00:47, 13.86it/s] Update steps: 93%|██████████████████████▍ | 9339/10000 [17:07<00:47, 13.87it/s] Update steps: 93%|██████████████████████▍ | 9341/10000 [17:07<00:47, 13.91it/s] Update steps: 93%|██████████████████████▍ | 9343/10000 [17:07<00:47, 13.90it/s] Update steps: 93%|██████████████████████▍ | 9345/10000 [17:07<00:47, 13.88it/s] Update steps: 93%|██████████████████████▍ | 9347/10000 [17:08<00:46, 13.93it/s] Update steps: 93%|██████████████████████▍ | 9349/10000 [17:08<00:47, 13.70it/s] Update steps: 94%|██████████████████████▍ | 9351/10000 [17:08<00:46, 13.82it/s] Update steps: 94%|██████████████████████▍ | 9353/10000 [17:08<00:46, 13.87it/s] Update steps: 94%|██████████████████████▍ | 9355/10000 [17:08<00:46, 13.91it/s] Update steps: 94%|██████████████████████▍ | 9357/10000 [17:08<00:46, 13.93it/s] Update steps: 94%|██████████████████████▍ | 9359/10000 [17:08<00:45, 13.95it/s] Update steps: 94%|██████████████████████▍ | 9361/10000 [17:09<00:45, 13.92it/s] Update steps: 94%|██████████████████████▍ | 9363/10000 [17:09<00:45, 13.87it/s] Update steps: 94%|██████████████████████▍ | 9365/10000 [17:09<00:46, 13.80it/s] Update steps: 94%|██████████████████████▍ | 9367/10000 [17:09<00:45, 13.78it/s] Update steps: 94%|██████████████████████▍ | 9369/10000 [17:09<00:45, 13.81it/s] Update steps: 94%|██████████████████████▍ | 9371/10000 [17:09<00:45, 13.90it/s] Update steps: 94%|██████████████████████▍ | 9373/10000 [17:09<00:44, 13.97it/s] Update steps: 94%|██████████████████████▌ | 9375/10000 [17:10<00:44, 13.99it/s] Update steps: 94%|██████████████████████▌ | 9377/10000 [17:10<00:44, 14.01it/s] Update steps: 94%|██████████████████████▌ | 9379/10000 [17:10<00:44, 13.97it/s] Update steps: 94%|██████████████████████▌ | 9381/10000 [17:10<00:44, 13.95it/s] Update steps: 94%|██████████████████████▌ | 9383/10000 [17:10<00:44, 13.96it/s] Update steps: 94%|██████████████████████▌ | 9385/10000 [17:10<00:43, 13.99it/s] Update steps: 94%|██████████████████████▌ | 9387/10000 [17:10<00:44, 13.92it/s] Update steps: 94%|██████████████████████▌ | 9389/10000 [17:11<00:44, 13.83it/s] Update steps: 94%|██████████████████████▌ | 9391/10000 [17:11<00:43, 13.92it/s] Update steps: 94%|██████████████████████▌ | 9393/10000 [17:11<00:43, 13.90it/s] Update steps: 94%|██████████████████████▌ | 9395/10000 [17:11<00:43, 13.95it/s] Update steps: 94%|██████████████████████▌ | 9397/10000 [17:11<00:43, 13.83it/s] Update steps: 94%|██████████████████████▌ | 9399/10000 [17:11<00:43, 13.84it/s] Update steps: 94%|██████████████████████▌ | 9401/10000 [17:11<00:43, 13.89it/s] Update steps: 94%|██████████████████████▌ | 9403/10000 [17:12<00:43, 13.87it/s] Update steps: 94%|██████████████████████▌ | 9405/10000 [17:12<00:42, 13.96it/s] Update steps: 94%|██████████████████████▌ | 9407/10000 [17:12<00:42, 13.97it/s] Update steps: 94%|██████████████████████▌ | 9409/10000 [17:12<00:42, 14.05it/s] Update steps: 94%|██████████████████████▌ | 9411/10000 [17:12<00:42, 13.98it/s] Update steps: 94%|██████████████████████▌ | 9413/10000 [17:12<00:42, 13.93it/s] Update steps: 94%|██████████████████████▌ | 9415/10000 [17:12<00:41, 13.98it/s] Update steps: 94%|██████████████████████▌ | 9417/10000 [17:13<00:41, 13.89it/s] Update steps: 94%|██████████████████████▌ | 9419/10000 [17:13<00:41, 13.96it/s] Update steps: 94%|██████████████████████▌ | 9421/10000 [17:13<00:41, 13.82it/s] Update steps: 94%|██████████████████████▌ | 9423/10000 [17:13<00:41, 13.89it/s] Update steps: 94%|██████████████████████▌ | 9425/10000 [17:13<00:41, 13.93it/s] Update steps: 94%|██████████████████████▌ | 9427/10000 [17:13<00:41, 13.91it/s] Update steps: 94%|██████████████████████▋ | 9429/10000 [17:13<00:41, 13.88it/s] Update steps: 94%|██████████████████████▋ | 9431/10000 [17:14<00:40, 13.93it/s] Update steps: 94%|██████████████████████▋ | 9433/10000 [17:14<00:40, 13.94it/s] Update steps: 94%|██████████████████████▋ | 9435/10000 [17:14<00:40, 13.84it/s] Update steps: 94%|██████████████████████▋ | 9437/10000 [17:14<00:40, 13.87it/s] Update steps: 94%|██████████████████████▋ | 9439/10000 [17:14<00:40, 13.86it/s] Update steps: 94%|██████████████████████▋ | 9441/10000 [17:14<00:40, 13.94it/s] Update steps: 94%|██████████████████████▋ | 9443/10000 [17:14<00:40, 13.90it/s] Update steps: 94%|██████████████████████▋ | 9445/10000 [17:15<00:39, 13.99it/s] Update steps: 94%|██████████████████████▋ | 9447/10000 [17:15<00:40, 13.67it/s] Update steps: 94%|██████████████████████▋ | 9449/10000 [17:15<00:39, 13.79it/s] Update steps: 95%|██████████████████████▋ | 9451/10000 [17:15<00:39, 13.84it/s] Update steps: 95%|██████████████████████▋ | 9453/10000 [17:15<00:39, 13.83it/s] Update steps: 95%|██████████████████████▋ | 9455/10000 [17:15<00:39, 13.82it/s] Update steps: 95%|██████████████████████▋ | 9457/10000 [17:15<00:39, 13.88it/s] Update steps: 95%|██████████████████████▋ | 9459/10000 [17:16<00:39, 13.85it/s] Update steps: 95%|██████████████████████▋ | 9461/10000 [17:16<00:39, 13.81it/s] Update steps: 95%|██████████████████████▋ | 9463/10000 [17:16<00:38, 13.91it/s] Update steps: 95%|██████████████████████▋ | 9465/10000 [17:16<00:38, 13.90it/s] Update steps: 95%|██████████████████████▋ | 9467/10000 [17:16<00:38, 13.77it/s] Update steps: 95%|██████████████████████▋ | 9469/10000 [17:16<00:38, 13.80it/s] Update steps: 95%|██████████████████████▋ | 9471/10000 [17:17<00:38, 13.83it/s] Update steps: 95%|██████████████████████▋ | 9473/10000 [17:17<00:38, 13.85it/s] Update steps: 95%|██████████████████████▋ | 9475/10000 [17:17<00:38, 13.75it/s] Update steps: 95%|██████████████████████▋ | 9477/10000 [17:17<00:38, 13.75it/s] Update steps: 95%|██████████████████████▋ | 9479/10000 [17:17<00:37, 13.77it/s] Update steps: 95%|██████████████████████▊ | 9481/10000 [17:17<00:37, 13.89it/s] Update steps: 95%|██████████████████████▊ | 9483/10000 [17:17<00:37, 13.89it/s] Update steps: 95%|██████████████████████▊ | 9485/10000 [17:18<00:36, 13.93it/s] Update steps: 95%|██████████████████████▊ | 9487/10000 [17:18<00:37, 13.78it/s] Update steps: 95%|██████████████████████▊ | 9489/10000 [17:18<00:37, 13.80it/s] Update steps: 95%|██████████████████████▊ | 9491/10000 [17:18<00:36, 13.76it/s] Update steps: 95%|██████████████████████▊ | 9493/10000 [17:18<00:36, 13.84it/s] Update steps: 95%|██████████████████████▊ | 9495/10000 [17:18<00:36, 13.78it/s] Update steps: 95%|██████████████████████▊ | 9497/10000 [17:18<00:36, 13.73it/s] Update steps: 95%|██████████████████████▊ | 9499/10000 [17:19<00:36, 13.87it/s] Update steps: 95%|██████████████████████▊ | 9501/10000 [17:19<00:35, 13.93it/s] Update steps: 95%|██████████████████████▊ | 9503/10000 [17:19<00:35, 13.97it/s] Update steps: 95%|██████████████████████▊ | 9505/10000 [17:19<00:35, 13.96it/s] Update steps: 95%|██████████████████████▊ | 9507/10000 [17:19<00:35, 13.92it/s] Update steps: 95%|██████████████████████▊ | 9509/10000 [17:19<00:35, 13.93it/s] Update steps: 95%|██████████████████████▊ | 9511/10000 [17:19<00:35, 13.93it/s] Update steps: 95%|██████████████████████▊ | 9513/10000 [17:20<00:34, 13.94it/s] Update steps: 95%|██████████████████████▊ | 9515/10000 [17:20<00:34, 13.97it/s] Update steps: 95%|██████████████████████▊ | 9517/10000 [17:20<00:34, 14.02it/s] Update steps: 95%|██████████████████████▊ | 9519/10000 [17:20<00:34, 14.04it/s] Update steps: 95%|██████████████████████▊ | 9521/10000 [17:20<00:35, 13.58it/s] Update steps: 95%|██████████████████████▊ | 9523/10000 [17:20<00:35, 13.58it/s] Update steps: 95%|██████████████████████▊ | 9525/10000 [17:20<00:34, 13.63it/s] Update steps: 95%|██████████████████████▊ | 9527/10000 [17:21<00:34, 13.65it/s] Update steps: 95%|██████████████████████▊ | 9529/10000 [17:21<00:34, 13.72it/s] Update steps: 95%|██████████████████████▊ | 9531/10000 [17:21<00:34, 13.76it/s] Update steps: 95%|██████████████████████▉ | 9533/10000 [17:21<00:33, 13.81it/s] Update steps: 95%|██████████████████████▉ | 9535/10000 [17:21<00:33, 13.90it/s] Update steps: 95%|██████████████████████▉ | 9537/10000 [17:21<00:33, 13.94it/s] Update steps: 95%|██████████████████████▉ | 9539/10000 [17:21<00:32, 14.00it/s] Update steps: 95%|██████████████████████▉ | 9541/10000 [17:22<00:32, 13.93it/s] Update steps: 95%|██████████████████████▉ | 9543/10000 [17:22<00:32, 13.87it/s] Update steps: 95%|██████████████████████▉ | 9545/10000 [17:22<00:32, 13.91it/s] Update steps: 95%|██████████████████████▉ | 9547/10000 [17:22<00:32, 13.96it/s] Update steps: 95%|██████████████████████▉ | 9549/10000 [17:22<00:32, 13.92it/s] Update steps: 96%|██████████████████████▉ | 9551/10000 [17:22<00:32, 13.97it/s] Update steps: 96%|██████████████████████▉ | 9553/10000 [17:22<00:32, 13.96it/s] Update steps: 96%|██████████████████████▉ | 9555/10000 [17:23<00:31, 13.96it/s] Update steps: 96%|██████████████████████▉ | 9557/10000 [17:23<00:32, 13.74it/s] Update steps: 96%|██████████████████████▉ | 9559/10000 [17:23<00:32, 13.75it/s] Update steps: 96%|██████████████████████▉ | 9561/10000 [17:23<00:31, 13.74it/s] Update steps: 96%|██████████████████████▉ | 9563/10000 [17:23<00:31, 13.78it/s] Update steps: 96%|██████████████████████▉ | 9565/10000 [17:23<00:31, 13.84it/s] Update steps: 96%|██████████████████████▉ | 9567/10000 [17:23<00:31, 13.63it/s] Update steps: 96%|██████████████████████▉ | 9569/10000 [17:24<00:31, 13.64it/s] Update steps: 96%|██████████████████████▉ | 9571/10000 [17:24<00:31, 13.76it/s] Update steps: 96%|██████████████████████▉ | 9573/10000 [17:24<00:31, 13.73it/s] Update steps: 96%|██████████████████████▉ | 9575/10000 [17:24<00:31, 13.68it/s] Update steps: 96%|██████████████████████▉ | 9577/10000 [17:24<00:30, 13.75it/s] Update steps: 96%|██████████████████████▉ | 9579/10000 [17:24<00:31, 13.27it/s] Update steps: 96%|██████████████████████▉ | 9581/10000 [17:24<00:31, 13.44it/s] Update steps: 96%|██████████████████████▉ | 9583/10000 [17:25<00:30, 13.50it/s] Update steps: 96%|███████████████████████ | 9585/10000 [17:25<00:30, 13.65it/s] Update steps: 96%|███████████████████████ | 9587/10000 [17:25<00:29, 13.79it/s] Update steps: 96%|███████████████████████ | 9589/10000 [17:25<00:29, 13.87it/s] Update steps: 96%|███████████████████████ | 9591/10000 [17:25<00:29, 13.86it/s] Update steps: 96%|███████████████████████ | 9593/10000 [17:25<00:29, 13.85it/s] Update steps: 96%|███████████████████████ | 9595/10000 [17:25<00:30, 13.50it/s] Update steps: 96%|███████████████████████ | 9597/10000 [17:26<00:29, 13.47it/s] Update steps: 96%|███████████████████████ | 9599/10000 [17:26<00:29, 13.54it/s] Update steps: 96%|███████████████████████ | 9601/10000 [17:26<00:29, 13.49it/s] Update steps: 96%|███████████████████████ | 9603/10000 [17:26<00:29, 13.61it/s] Update steps: 96%|███████████████████████ | 9605/10000 [17:26<00:28, 13.70it/s] Update steps: 96%|███████████████████████ | 9607/10000 [17:26<00:28, 13.79it/s] Update steps: 96%|███████████████████████ | 9609/10000 [17:27<00:28, 13.79it/s] Update steps: 96%|███████████████████████ | 9611/10000 [17:27<00:28, 13.86it/s] Update steps: 96%|███████████████████████ | 9613/10000 [17:27<00:27, 13.87it/s] Update steps: 96%|███████████████████████ | 9615/10000 [17:27<00:27, 13.95it/s] Update steps: 96%|███████████████████████ | 9617/10000 [17:27<00:27, 13.95it/s] Update steps: 96%|███████████████████████ | 9619/10000 [17:27<00:27, 13.93it/s] Update steps: 96%|███████████████████████ | 9621/10000 [17:27<00:27, 13.90it/s] Update steps: 96%|███████████████████████ | 9623/10000 [17:28<00:27, 13.58it/s] Update steps: 96%|███████████████████████ | 9625/10000 [17:28<00:28, 13.39it/s] Update steps: 96%|███████████████████████ | 9627/10000 [17:28<00:27, 13.58it/s] Update steps: 96%|███████████████████████ | 9629/10000 [17:28<00:27, 13.59it/s] Update steps: 96%|███████████████████████ | 9631/10000 [17:28<00:27, 13.64it/s] Update steps: 96%|███████████████████████ | 9633/10000 [17:28<00:26, 13.73it/s] Update steps: 96%|███████████████████████ | 9635/10000 [17:28<00:26, 13.82it/s] Update steps: 96%|███████████████████████▏| 9637/10000 [17:29<00:26, 13.89it/s] Update steps: 96%|███████████████████████▏| 9639/10000 [17:29<00:25, 13.92it/s] Update steps: 96%|███████████████████████▏| 9641/10000 [17:29<00:25, 13.95it/s] Update steps: 96%|███████████████████████▏| 9643/10000 [17:29<00:25, 14.02it/s] Update steps: 96%|███████████████████████▏| 9645/10000 [17:29<00:25, 14.03it/s] Update steps: 96%|███████████████████████▏| 9647/10000 [17:29<00:25, 14.07it/s] Update steps: 96%|███████████████████████▏| 9649/10000 [17:29<00:24, 14.05it/s] Update steps: 97%|███████████████████████▏| 9651/10000 [17:30<00:24, 14.00it/s] Update steps: 97%|███████████████████████▏| 9653/10000 [17:30<00:24, 13.95it/s] Update steps: 97%|███████████████████████▏| 9655/10000 [17:30<00:24, 13.98it/s] Update steps: 97%|███████████████████████▏| 9657/10000 [17:30<00:24, 13.98it/s] Update steps: 97%|███████████████████████▏| 9659/10000 [17:30<00:24, 14.01it/s] Update steps: 97%|███████████████████████▏| 9661/10000 [17:30<00:24, 13.93it/s] Update steps: 97%|███████████████████████▏| 9663/10000 [17:30<00:24, 13.92it/s] Update steps: 97%|███████████████████████▏| 9665/10000 [17:31<00:23, 13.96it/s] Update steps: 97%|███████████████████████▏| 9667/10000 [17:31<00:23, 13.96it/s] Update steps: 97%|███████████████████████▏| 9669/10000 [17:31<00:23, 13.95it/s] Update steps: 97%|███████████████████████▏| 9671/10000 [17:31<00:23, 13.90it/s] Update steps: 97%|███████████████████████▏| 9673/10000 [17:31<00:23, 13.90it/s] Update steps: 97%|███████████████████████▏| 9675/10000 [17:31<00:23, 13.95it/s] Update steps: 97%|███████████████████████▏| 9677/10000 [17:31<00:23, 13.97it/s] Update steps: 97%|███████████████████████▏| 9679/10000 [17:32<00:22, 13.96it/s] Update steps: 97%|███████████████████████▏| 9681/10000 [17:32<00:23, 13.85it/s] Update steps: 97%|███████████████████████▏| 9683/10000 [17:32<00:22, 13.89it/s] Update steps: 97%|███████████████████████▏| 9685/10000 [17:32<00:22, 13.97it/s] Update steps: 97%|███████████████████████▏| 9687/10000 [17:32<00:22, 13.97it/s] Update steps: 97%|███████████████████████▎| 9689/10000 [17:32<00:22, 14.03it/s] Update steps: 97%|███████████████████████▎| 9691/10000 [17:32<00:22, 13.99it/s] Update steps: 97%|███████████████████████▎| 9693/10000 [17:33<00:22, 13.85it/s] Update steps: 97%|███████████████████████▎| 9695/10000 [17:33<00:21, 13.90it/s] Update steps: 97%|███████████████████████▎| 9697/10000 [17:33<00:21, 13.85it/s] Update steps: 97%|███████████████████████▎| 9699/10000 [17:33<00:22, 13.57it/s] Update steps: 97%|███████████████████████▎| 9701/10000 [17:33<00:21, 13.70it/s] Update steps: 97%|███████████████████████▎| 9703/10000 [17:33<00:21, 13.80it/s] Update steps: 97%|███████████████████████▎| 9705/10000 [17:33<00:21, 13.81it/s] Update steps: 97%|███████████████████████▎| 9707/10000 [17:34<00:21, 13.91it/s] Update steps: 97%|███████████████████████▎| 9709/10000 [17:34<00:20, 13.97it/s] Update steps: 97%|███████████████████████▎| 9711/10000 [17:34<00:20, 13.91it/s] Update steps: 97%|███████████████████████▎| 9713/10000 [17:34<00:20, 13.89it/s] Update steps: 97%|███████████████████████▎| 9715/10000 [17:34<00:20, 13.93it/s] Update steps: 97%|███████████████████████▎| 9717/10000 [17:34<00:20, 13.83it/s] Update steps: 97%|███████████████████████▎| 9719/10000 [17:34<00:20, 13.74it/s] Update steps: 97%|███████████████████████▎| 9721/10000 [17:35<00:20, 13.82it/s] Update steps: 97%|███████████████████████▎| 9723/10000 [17:35<00:19, 13.89it/s] Update steps: 97%|███████████████████████▎| 9725/10000 [17:35<00:19, 13.97it/s] Update steps: 97%|███████████████████████▎| 9727/10000 [17:35<00:19, 14.00it/s] Update steps: 97%|███████████████████████▎| 9729/10000 [17:35<00:19, 13.96it/s] Update steps: 97%|███████████████████████▎| 9731/10000 [17:35<00:19, 14.04it/s] Update steps: 97%|███████████████████████▎| 9733/10000 [17:35<00:19, 14.03it/s] Update steps: 97%|███████████████████████▎| 9735/10000 [17:36<00:18, 14.00it/s] Update steps: 97%|███████████████████████▎| 9737/10000 [17:36<00:19, 13.75it/s] Update steps: 97%|███████████████████████▎| 9739/10000 [17:36<00:18, 13.80it/s] Update steps: 97%|███████████████████████▍| 9741/10000 [17:36<00:18, 13.83it/s] Update steps: 97%|███████████████████████▍| 9743/10000 [17:36<00:18, 13.78it/s] Update steps: 97%|███████████████████████▍| 9745/10000 [17:36<00:18, 13.81it/s] Update steps: 97%|███████████████████████▍| 9747/10000 [17:36<00:18, 13.86it/s] Update steps: 97%|███████████████████████▍| 9749/10000 [17:37<00:18, 13.88it/s] Update steps: 98%|███████████████████████▍| 9751/10000 [17:37<00:18, 13.35it/s] Update steps: 98%|███████████████████████▍| 9753/10000 [17:37<00:18, 13.51it/s] Update steps: 98%|███████████████████████▍| 9755/10000 [17:37<00:18, 13.53it/s] Update steps: 98%|███████████████████████▍| 9757/10000 [17:37<00:17, 13.68it/s] Update steps: 98%|███████████████████████▍| 9759/10000 [17:37<00:17, 13.73it/s] Update steps: 98%|███████████████████████▍| 9761/10000 [17:37<00:17, 13.84it/s] Update steps: 98%|███████████████████████▍| 9763/10000 [17:38<00:17, 13.84it/s] Update steps: 98%|███████████████████████▍| 9765/10000 [17:38<00:16, 13.84it/s] Update steps: 98%|███████████████████████▍| 9767/10000 [17:38<00:16, 13.88it/s] Update steps: 98%|███████████████████████▍| 9769/10000 [17:38<00:16, 13.96it/s] Update steps: 98%|███████████████████████▍| 9771/10000 [17:38<00:16, 13.98it/s] Update steps: 98%|███████████████████████▍| 9773/10000 [17:38<00:16, 13.93it/s] Update steps: 98%|███████████████████████▍| 9775/10000 [17:38<00:16, 13.84it/s] Update steps: 98%|███████████████████████▍| 9777/10000 [17:39<00:16, 13.91it/s] Update steps: 98%|███████████████████████▍| 9779/10000 [17:39<00:16, 13.66it/s] Update steps: 98%|███████████████████████▍| 9781/10000 [17:39<00:15, 13.70it/s] Update steps: 98%|███████████████████████▍| 9783/10000 [17:39<00:15, 13.70it/s] Update steps: 98%|███████████████████████▍| 9785/10000 [17:39<00:15, 13.68it/s] Update steps: 98%|███████████████████████▍| 9787/10000 [17:39<00:15, 13.70it/s] Update steps: 98%|███████████████████████▍| 9789/10000 [17:40<00:15, 13.76it/s] Update steps: 98%|███████████████████████▍| 9791/10000 [17:40<00:15, 13.74it/s] Update steps: 98%|███████████████████████▌| 9793/10000 [17:40<00:15, 13.67it/s] Update steps: 98%|███████████████████████▌| 9795/10000 [17:40<00:14, 13.81it/s] Update steps: 98%|███████████████████████▌| 9797/10000 [17:40<00:14, 13.77it/s] Update steps: 98%|███████████████████████▌| 9799/10000 [17:40<00:14, 13.82it/s] Update steps: 98%|███████████████████████▌| 9801/10000 [17:40<00:14, 13.85it/s] Update steps: 98%|███████████████████████▌| 9803/10000 [17:41<00:14, 13.92it/s] Update steps: 98%|███████████████████████▌| 9805/10000 [17:41<00:14, 13.86it/s] Update steps: 98%|███████████████████████▌| 9807/10000 [17:41<00:14, 13.60it/s] Update steps: 98%|███████████████████████▌| 9809/10000 [17:41<00:14, 13.64it/s] Update steps: 98%|███████████████████████▌| 9811/10000 [17:41<00:13, 13.77it/s] Update steps: 98%|███████████████████████▌| 9813/10000 [17:41<00:13, 13.85it/s] Update steps: 98%|███████████████████████▌| 9815/10000 [17:41<00:13, 13.89it/s] Update steps: 98%|███████████████████████▌| 9817/10000 [17:42<00:13, 13.92it/s] Update steps: 98%|███████████████████████▌| 9819/10000 [17:42<00:13, 13.92it/s] Update steps: 98%|███████████████████████▌| 9821/10000 [17:42<00:12, 13.87it/s] Update steps: 98%|███████████████████████▌| 9823/10000 [17:42<00:12, 13.87it/s] Update steps: 98%|███████████████████████▌| 9825/10000 [17:42<00:12, 13.86it/s] Update steps: 98%|███████████████████████▌| 9827/10000 [17:42<00:12, 13.88it/s] Update steps: 98%|███████████████████████▌| 9829/10000 [17:42<00:12, 13.88it/s] Update steps: 98%|███████████████████████▌| 9831/10000 [17:43<00:12, 13.94it/s] Update steps: 98%|███████████████████████▌| 9833/10000 [17:43<00:11, 13.97it/s] Update steps: 98%|███████████████████████▌| 9835/10000 [17:43<00:11, 13.93it/s] Update steps: 98%|███████████████████████▌| 9837/10000 [17:43<00:11, 13.85it/s] Update steps: 98%|███████████████████████▌| 9839/10000 [17:43<00:11, 13.72it/s] Update steps: 98%|███████████████████████▌| 9841/10000 [17:43<00:11, 13.81it/s] Update steps: 98%|███████████████████████▌| 9843/10000 [17:43<00:11, 13.90it/s] Update steps: 98%|███████████████████████▋| 9845/10000 [17:44<00:11, 13.93it/s] Update steps: 98%|███████████████████████▋| 9847/10000 [17:44<00:11, 13.45it/s] Update steps: 98%|███████████████████████▋| 9849/10000 [17:44<00:11, 13.48it/s] Update steps: 99%|███████████████████████▋| 9851/10000 [17:44<00:11, 13.52it/s] Update steps: 99%|███████████████████████▋| 9853/10000 [17:44<00:10, 13.69it/s] Update steps: 99%|███████████████████████▋| 9855/10000 [17:44<00:10, 13.76it/s] Update steps: 99%|███████████████████████▋| 9857/10000 [17:44<00:10, 13.81it/s] Update steps: 99%|███████████████████████▋| 9859/10000 [17:45<00:10, 13.89it/s] Update steps: 99%|███████████████████████▋| 9861/10000 [17:45<00:10, 13.83it/s] Update steps: 99%|███████████████████████▋| 9863/10000 [17:45<00:09, 13.78it/s] Update steps: 99%|███████████████████████▋| 9865/10000 [17:45<00:09, 13.87it/s] Update steps: 99%|███████████████████████▋| 9867/10000 [17:45<00:09, 13.89it/s] Update steps: 99%|███████████████████████▋| 9869/10000 [17:45<00:09, 13.81it/s] Update steps: 99%|███████████████████████▋| 9871/10000 [17:45<00:09, 13.74it/s] Update steps: 99%|███████████████████████▋| 9873/10000 [17:46<00:09, 13.80it/s] Update steps: 99%|███████████████████████▋| 9875/10000 [17:46<00:09, 13.88it/s] Update steps: 99%|███████████████████████▋| 9877/10000 [17:46<00:08, 13.93it/s] Update steps: 99%|███████████████████████▋| 9879/10000 [17:46<00:08, 13.98it/s] Update steps: 99%|███████████████████████▋| 9881/10000 [17:46<00:08, 13.96it/s] Update steps: 99%|███████████████████████▋| 9883/10000 [17:46<00:08, 14.00it/s] Update steps: 99%|███████████████████████▋| 9885/10000 [17:46<00:08, 14.03it/s] Update steps: 99%|███████████████████████▋| 9887/10000 [17:47<00:08, 14.05it/s] Update steps: 99%|███████████████████████▋| 9889/10000 [17:47<00:07, 13.95it/s] Update steps: 99%|███████████████████████▋| 9891/10000 [17:47<00:07, 13.81it/s] Update steps: 99%|███████████████████████▋| 9893/10000 [17:47<00:07, 13.75it/s] Update steps: 99%|███████████████████████▋| 9895/10000 [17:47<00:07, 13.80it/s] Update steps: 99%|███████████████████████▊| 9897/10000 [17:47<00:07, 13.66it/s] Update steps: 99%|███████████████████████▊| 9899/10000 [17:47<00:07, 13.80it/s] Update steps: 99%|███████████████████████▊| 9901/10000 [17:48<00:07, 13.80it/s] Update steps: 99%|███████████████████████▊| 9903/10000 [17:48<00:07, 13.84it/s] Update steps: 99%|███████████████████████▊| 9905/10000 [17:48<00:06, 13.95it/s] Update steps: 99%|███████████████████████▊| 9907/10000 [17:48<00:06, 13.94it/s] Update steps: 99%|███████████████████████▊| 9909/10000 [17:48<00:06, 13.93it/s] Update steps: 99%|███████████████████████▊| 9911/10000 [17:48<00:06, 13.79it/s] Update steps: 99%|███████████████████████▊| 9913/10000 [17:48<00:06, 13.85it/s] Update steps: 99%|███████████████████████▊| 9915/10000 [17:49<00:06, 13.91it/s] Update steps: 99%|███████████████████████▊| 9917/10000 [17:49<00:05, 13.97it/s] Update steps: 99%|███████████████████████▊| 9919/10000 [17:49<00:05, 14.01it/s] Update steps: 99%|███████████████████████▊| 9921/10000 [17:49<00:05, 14.04it/s] Update steps: 99%|███████████████████████▊| 9923/10000 [17:49<00:05, 13.79it/s] Update steps: 99%|███████████████████████▊| 9925/10000 [17:49<00:05, 13.88it/s] Update steps: 99%|███████████████████████▊| 9927/10000 [17:49<00:05, 13.93it/s] Update steps: 99%|███████████████████████▊| 9929/10000 [17:50<00:05, 13.91it/s] Update steps: 99%|███████████████████████▊| 9931/10000 [17:50<00:04, 13.90it/s] Update steps: 99%|███████████████████████▊| 9933/10000 [17:50<00:04, 13.96it/s] Update steps: 99%|███████████████████████▊| 9935/10000 [17:50<00:04, 13.96it/s] Update steps: 99%|███████████████████████▊| 9937/10000 [17:50<00:04, 13.95it/s] Update steps: 99%|███████████████████████▊| 9939/10000 [17:50<00:04, 14.00it/s] Update steps: 99%|███████████████████████▊| 9941/10000 [17:50<00:04, 14.02it/s] Update steps: 99%|███████████████████████▊| 9943/10000 [17:51<00:04, 13.93it/s] Update steps: 99%|███████████████████████▊| 9945/10000 [17:51<00:04, 13.75it/s] Update steps: 99%|███████████████████████▊| 9947/10000 [17:51<00:03, 13.72it/s] Update steps: 99%|███████████████████████▉| 9949/10000 [17:51<00:03, 13.84it/s] Update steps: 100%|███████████████████████▉| 9951/10000 [17:51<00:03, 13.90it/s] Update steps: 100%|███████████████████████▉| 9953/10000 [17:51<00:03, 13.80it/s] Update steps: 100%|███████████████████████▉| 9955/10000 [17:52<00:03, 13.47it/s] Update steps: 100%|███████████████████████▉| 9957/10000 [17:52<00:03, 13.61it/s] Update steps: 100%|███████████████████████▉| 9959/10000 [17:52<00:02, 13.70it/s] Update steps: 100%|███████████████████████▉| 9961/10000 [17:52<00:02, 13.65it/s] Update steps: 100%|███████████████████████▉| 9963/10000 [17:52<00:02, 13.66it/s] Update steps: 100%|███████████████████████▉| 9965/10000 [17:52<00:02, 13.77it/s] Update steps: 100%|███████████████████████▉| 9967/10000 [17:52<00:02, 13.83it/s] Update steps: 100%|███████████████████████▉| 9969/10000 [17:53<00:02, 13.91it/s] Update steps: 100%|███████████████████████▉| 9971/10000 [17:53<00:02, 13.85it/s] Update steps: 100%|███████████████████████▉| 9973/10000 [17:53<00:01, 13.87it/s] Update steps: 100%|███████████████████████▉| 9975/10000 [17:53<00:01, 13.84it/s] Update steps: 100%|███████████████████████▉| 9977/10000 [17:53<00:01, 13.54it/s] Update steps: 100%|███████████████████████▉| 9979/10000 [17:53<00:01, 13.62it/s] Update steps: 100%|███████████████████████▉| 9981/10000 [17:53<00:01, 13.69it/s] Update steps: 100%|███████████████████████▉| 9983/10000 [17:54<00:01, 13.73it/s] Update steps: 100%|███████████████████████▉| 9985/10000 [17:54<00:01, 13.77it/s] Update steps: 100%|███████████████████████▉| 9987/10000 [17:54<00:00, 13.87it/s] Update steps: 100%|███████████████████████▉| 9989/10000 [17:54<00:00, 13.94it/s] Update steps: 100%|███████████████████████▉| 9991/10000 [17:54<00:00, 13.93it/s] Update steps: 100%|███████████████████████▉| 9993/10000 [17:54<00:00, 13.95it/s] Update steps: 100%|███████████████████████▉| 9995/10000 [17:54<00:00, 13.84it/s] Update steps: 100%|███████████████████████▉| 9997/10000 [17:55<00:00, 13.81it/s] Update steps: 100%|███████████████████████▉| 9999/10000 [17:55<00:00, 13.85it/s]2026-04-17 11:11:15.840 | INFO | __main__:main:452 - Saving model and optimizer to /mnt/dhwfile/tancheng/panjiabao/Result/SAC_C4/work_dirs/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_20260417_105249/model_10000, update step 10000 +saving model weight without quantized layer +2026-04-17 11:11:17.157 | INFO | __main__:main:490 - Performing evaluation at step 10000 +2026-04-17 11:11:17.158 | INFO | utils.eval:evaluate_model:21 - Loading validation data from 8 local validation files +2026-04-17 11:11:17.901 | INFO | utils.eval:evaluate_model:43 - Loaded validation dataset in 0.74 seconds +2026-04-17 11:11:17.902 | INFO | utils.eval:evaluate_model:60 - Eval set prepared in 0.74 seconds + Update steps: 100%|███████████████████████| 10000/10000 [18:06<00:00, 13.85it/s]2026-04-17 11:12:02.410 | INFO | __main__:main:505 - Eval loss at step 10000: 3.403851938099257, Eval perplexity: 30.079742486752227 +2026-04-17 11:12:02.417 | INFO | __main__:main:405 - Reached max number of update steps (10000). Stopping training. +2026-04-17 11:12:02.417 | INFO | __main__:main:406 - Rank 0 stopping training. +2026-04-17 11:12:02.417 | INFO | __main__:main:539 - Training finished + Update steps: 100%|███████████████████████| 10000/10000 [18:42<00:00, 8.91it/s] +2026-04-17 11:12:02.419 | INFO | __main__:main:574 - Running final evaluation +2026-04-17 11:12:02.793 | INFO | utils.eval:evaluate_model:21 - Loading validation data from 8 local validation files +2026-04-17 11:12:03.645 | INFO | utils.eval:evaluate_model:43 - Loaded validation dataset in 0.85 seconds +2026-04-17 11:12:03.647 | INFO | utils.eval:evaluate_model:60 - Eval set prepared in 0.86 seconds +Rank 1 finished successfullyRank 2 finished successfully + +Rank 3 finished successfully +2026-04-17 11:12:39.027 | INFO | __main__:main:594 - Final eval loss: 3.403851938099257, Final eval perplexity: 30.079742486752227 +2026-04-17 11:12:39.027 | INFO | __main__:main:596 - Script finished successfully +Rank 0 finished successfully +wandb: +wandb: You can sync this run to the cloud by running: +wandb: wandb sync /mnt/dhwfile/tancheng/panjiabao/Result/SAC_C4/work_dirs/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_20260417_105249/wandb/offline-run-20260417_105314-6uk0fpmn +wandb: Find logs at: ../../../../dhwfile/tancheng/panjiabao/Result/SAC_C4/work_dirs/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_20260417_105249/wandb/offline-run-20260417_105314-6uk0fpmn/logs +[rank0]:[W417 11:12:40.376287478 ProcessGroupNCCL.cpp:1250] Warning: WARNING: process group has NOT been destroyed before we destruct ProcessGroupNCCL. On normal program exit, the application should call destroy_process_group to ensure that any pending NCCL operations have finished in this process. In rare cases this process can exit before this point and block the progress of another member of the process group. This constraint has always been present, but this warning has only been added since PyTorch 2.4 (function operator()) diff --git a/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_A100_ppl_30_0797_20260417_105249/model_10000/config.json b/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_A100_ppl_30_0797_20260417_105249/model_10000/config.json new file mode 100644 index 0000000000000000000000000000000000000000..3da42b2ddfc762898ec6f30180447acb6dcd2e46 --- /dev/null +++ b/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_A100_ppl_30_0797_20260417_105249/model_10000/config.json @@ -0,0 +1,31 @@ +{ + "architectures": [ + "LlamaForCausalLM" + ], + "attention_bias": false, + "attention_dropout": 0.0, + "bos_token_id": 0, + "dtype": "bfloat16", + "eos_token_id": 1, + "head_dim": 64, + "hidden_act": "silu", + "hidden_size": 512, + "initializer_range": 0.02, + "intermediate_size": 1376, + "max_position_embeddings": 2048, + "max_sequence_length": 1024, + "mlp_bias": false, + "model_type": "llama", + "num_attention_heads": 8, + "num_hidden_layers": 8, + "num_key_value_heads": 8, + "pad_token_id": -1, + "pretraining_tp": 1, + "rms_norm_eps": 1e-06, + "rope_scaling": null, + "rope_theta": 10000.0, + "tie_word_embeddings": false, + "transformers_version": "4.57.3", + "use_cache": true, + "vocab_size": 32000 +} diff --git a/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_A100_ppl_30_0797_20260417_105249/model_10000/model.safetensors b/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_A100_ppl_30_0797_20260417_105249/model_10000/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..a9a103604f3cced1aab7a1bb3c060efffb92dc62 --- /dev/null +++ b/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_A100_ppl_30_0797_20260417_105249/model_10000/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:74aeaeb75678d0eec8eb94e9e0861ca32d7ced2933a62412a356d2689ad77887 +size 116156896 diff --git a/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_A100_ppl_30_0797_20260417_105249/model_10000/optimizer.pt b/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_A100_ppl_30_0797_20260417_105249/model_10000/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..338dcba8a160bf9e0a61f95b0a9aa525792423bc --- /dev/null +++ b/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_A100_ppl_30_0797_20260417_105249/model_10000/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:56f346becd55f4e2001d3258634c49e924275330a471cd7623ddf0f4f8ceaf7e +size 232362810 diff --git a/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_A100_ppl_30_0797_20260417_105249/model_10000/pytorch_model.bin b/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_A100_ppl_30_0797_20260417_105249/model_10000/pytorch_model.bin new file mode 100644 index 0000000000000000000000000000000000000000..e8d13e51c5572651f1f02ce777a8947285a00dce --- /dev/null +++ b/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_A100_ppl_30_0797_20260417_105249/model_10000/pytorch_model.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:95027b6442a00f34f1a766bf3a009d02d7f354d62667f26b528e7d9f72d43129 +size 116181735 diff --git a/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_A100_ppl_30_0797_20260417_105249/model_10000/training_state.json b/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_A100_ppl_30_0797_20260417_105249/model_10000/training_state.json new file mode 100644 index 0000000000000000000000000000000000000000..5177b3b9c2f9a8f52264b496905a19c1db389a95 --- /dev/null +++ b/60m/adamw_lr3e_3_b1_0_9_b2_0_99_eps_1e_8_A100_ppl_30_0797_20260417_105249/model_10000/training_state.json @@ -0,0 +1,7 @@ +{ + "global_step": 10000, + "update_step": 10000, + "tokens_seen": 999923888, + "tokens_seen_before": 999826428, + "update_time": 0.25369739532470703 +} \ No newline at end of file