decoding_results/fast_beam_search/log-decode-epoch-40-avg-10-beam-20.0-max-contexts-8-max-states-64-blank-penalty-0.0-use-averaged-model-2024-01-16-10-04-04

2024-01-16 10:04:04,359 INFO [decode.py:673] Decoding started
2024-01-16 10:04:04,359 INFO [decode.py:679] Device: cuda:0
2024-01-16 10:04:04,423 INFO [lexicon.py:168] Loading pre-compiled data/lang_bbpe_500/Linv.pt
2024-01-16 10:04:04,436 INFO [decode.py:691] {'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 50, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'warm_step': 2000, 'env_info': {'k2-version': '1.24.4', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': '2989b0b1186fa6022932804f5b39fbb2781ebf42', 'k2-git-date': 'Fri Nov 24 11:34:10 2023', 'lhotse-version': '1.19.0.dev+git.d1ae9c05.dirty', 'torch-version': '1.11.0+cu102', 'torch-cuda-available': True, 'torch-cuda-version': '10.2', 'python-version': '3.9', 'icefall-git-branch': 'dev/aishell-zipformer-bbpe', 'icefall-git-sha1': 'bce81394-clean', 'icefall-git-date': 'Thu Jan 11 09:56:01 2024', 'icefall-path': '/star-home/jinzengrui/lib/miniconda3/envs/dev39/lib/python3.9/site-packages/icefall-1.0-py3.9.egg', 'k2-path': '/star-home/jinzengrui/lib/miniconda3/envs/dev39/lib/python3.9/site-packages/k2-1.24.4.dev20231207+cuda10.2.torch1.11.0-py3.9-linux-x86_64.egg/k2/__init__.py', 'lhotse-path': '/star-home/jinzengrui/lib/miniconda3/envs/dev39/lib/python3.9/site-packages/lhotse-1.19.0.dev0+git.d1ae9c05.dirty-py3.9.egg/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-1-1207150822-75498b8c5f-55j4z', 'IP address': '10.177.74.211'}, 'epoch': 40, 'iter': 0, 'avg': 10, 'use_averaged_model': True, 'exp_dir': PosixPath('zipformer_bbpe/exp-context-size-2-lr-epochs-10-spec-aug-20-disable-musan'), 'bpe_model': 'data/lang_bbpe_500/bbpe.model', 'lang_dir': PosixPath('data/lang_bbpe_500'), 'decoding_method': 'fast_beam_search', 'beam_size': 4, 'beam': 20.0, 'ngram_lm_scale': 0.01, 'ilme_scale': 0.2, 'max_contexts': 8, 'max_states': 64, 'context_size': 2, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, 'blank_penalty': 0.0, 'num_encoder_layers': '2,2,3,4,3,2', 'downsampling_factor': '1,2,4,8,4,2', 'feedforward_dim': '512,768,1024,1536,1024,768', 'num_heads': '4,4,4,8,4,4', 'encoder_dim': '192,256,384,512,384,256', 'query_head_dim': '32', 'value_head_dim': '12', 'pos_head_dim': '4', 'pos_dim': 48, 'encoder_unmasked_dim': '192,192,256,256,256,192', 'cnn_module_kernel': '31,31,15,15,15,31', 'decoder_dim': 512, 'joiner_dim': 512, 'causal': False, 'chunk_size': '16,32,64,-1', 'left_context_frames': '64,128,256,-1', 'manifest_dir': PosixPath('data/fbank'), 'max_duration': 200.0, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'on_the_fly_feats': False, 'shuffle': True, 'drop_last': True, 'return_cuts': True, 'num_workers': 2, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'res_dir': PosixPath('zipformer_bbpe/exp-context-size-2-lr-epochs-10-spec-aug-20-disable-musan/fast_beam_search'), 'suffix': 'epoch-40-avg-10-beam-20.0-max-contexts-8-max-states-64-blank-penalty-0.0-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
2024-01-16 10:04:04,436 INFO [decode.py:693] About to create model
2024-01-16 10:04:04,956 INFO [decode.py:760] Calculating the averaged model over epoch range from 30 (excluded) to 40
2024-01-16 10:04:12,532 INFO [decode.py:791] Number of model parameters: 65549011
2024-01-16 10:04:12,532 INFO [asr_datamodule.py:371] About to get dev cuts
2024-01-16 10:04:12,549 INFO [asr_datamodule.py:312] About to create dev dataset
2024-01-16 10:04:13,059 INFO [asr_datamodule.py:329] About to create dev dataloader
2024-01-16 10:04:13,059 INFO [asr_datamodule.py:376] About to get test cuts
2024-01-16 10:04:13,061 INFO [asr_datamodule.py:341] About to create test dataset
2024-01-16 10:04:15,062 INFO [decode.py:564] batch 0/?, cuts processed until now is 35
2024-01-16 10:04:27,318 INFO [decode.py:564] batch 20/?, cuts processed until now is 785
2024-01-16 10:04:39,148 INFO [decode.py:564] batch 40/?, cuts processed until now is 1572
2024-01-16 10:04:50,338 INFO [decode.py:564] batch 60/?, cuts processed until now is 2433
2024-01-16 10:05:03,143 INFO [decode.py:564] batch 80/?, cuts processed until now is 3154
2024-01-16 10:05:11,212 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([1.7125, 2.7740, 2.7712, 2.0904], device='cuda:0')
2024-01-16 10:05:14,032 INFO [decode.py:564] batch 100/?, cuts processed until now is 4023
2024-01-16 10:05:26,372 INFO [decode.py:564] batch 120/?, cuts processed until now is 4776
2024-01-16 10:05:37,471 INFO [decode.py:564] batch 140/?, cuts processed until now is 5662
2024-01-16 10:05:39,479 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([1.1403, 1.6074, 1.8219, 3.2897], device='cuda:0')
2024-01-16 10:05:45,095 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([1.1947, 1.4890, 1.7581, 3.0378], device='cuda:0')
2024-01-16 10:05:48,552 INFO [decode.py:564] batch 160/?, cuts processed until now is 6552
2024-01-16 10:05:59,780 INFO [decode.py:564] batch 180/?, cuts processed until now is 7386
2024-01-16 10:06:11,198 INFO [decode.py:564] batch 200/?, cuts processed until now is 8204
2024-01-16 10:06:11,753 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([4.3423, 3.6335, 3.9994, 3.4270], device='cuda:0')
2024-01-16 10:06:23,159 INFO [decode.py:564] batch 220/?, cuts processed until now is 8999
2024-01-16 10:06:36,041 INFO [decode.py:564] batch 240/?, cuts processed until now is 9811
2024-01-16 10:06:47,892 INFO [decode.py:564] batch 260/?, cuts processed until now is 10620
2024-01-16 10:06:58,301 INFO [decode.py:564] batch 280/?, cuts processed until now is 11537
2024-01-16 10:07:10,090 INFO [decode.py:564] batch 300/?, cuts processed until now is 12391
2024-01-16 10:07:21,683 INFO [decode.py:564] batch 320/?, cuts processed until now is 13152
2024-01-16 10:07:33,975 INFO [decode.py:564] batch 340/?, cuts processed until now is 13879
2024-01-16 10:07:45,332 INFO [decode.py:580] The transcripts are stored in zipformer_bbpe/exp-context-size-2-lr-epochs-10-spec-aug-20-disable-musan/fast_beam_search/recogs-dev-blank_penalty_0.0_beam_20.0_max_contexts_8_max_states_64-epoch-40-avg-10-beam-20.0-max-contexts-8-max-states-64-blank-penalty-0.0-use-averaged-model.txt
2024-01-16 10:07:45,825 INFO [utils.py:565] [dev-blank_penalty_0.0_beam_20.0_max_contexts_8_max_states_64] %WER 4.17% [8566 / 205341, 197 ins, 500 del, 7869 sub ]
2024-01-16 10:07:46,507 INFO [decode.py:598] Wrote detailed error stats to zipformer_bbpe/exp-context-size-2-lr-epochs-10-spec-aug-20-disable-musan/fast_beam_search/errs-dev-blank_penalty_0.0_beam_20.0_max_contexts_8_max_states_64-epoch-40-avg-10-beam-20.0-max-contexts-8-max-states-64-blank-penalty-0.0-use-averaged-model.txt
2024-01-16 10:07:46,527 INFO [decode.py:614] 
For dev, WER of different settings are:
blank_penalty_0.0_beam_20.0_max_contexts_8_max_states_64	4.17	best for dev

2024-01-16 10:07:48,132 INFO [decode.py:564] batch 0/?, cuts processed until now is 31
2024-01-16 10:08:04,342 INFO [decode.py:564] batch 20/?, cuts processed until now is 697
2024-01-16 10:08:17,927 INFO [decode.py:564] batch 40/?, cuts processed until now is 1395
2024-01-16 10:08:29,163 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.4244, 2.1965, 2.9948, 2.2683, 2.6988, 2.7314, 2.9246, 2.6597],
       device='cuda:0')
2024-01-16 10:08:29,706 INFO [decode.py:564] batch 60/?, cuts processed until now is 2169
2024-01-16 10:08:33,536 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([3.5973, 3.2926, 2.6284, 2.3168], device='cuda:0')
2024-01-16 10:08:43,311 INFO [decode.py:564] batch 80/?, cuts processed until now is 2801
2024-01-16 10:08:55,694 INFO [decode.py:564] batch 100/?, cuts processed until now is 3578
2024-01-16 10:09:09,070 INFO [decode.py:564] batch 120/?, cuts processed until now is 4249
2024-01-16 10:09:20,752 INFO [decode.py:564] batch 140/?, cuts processed until now is 5061
2024-01-16 10:09:27,732 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([5.9109, 5.6864, 5.6041, 5.6176], device='cuda:0')
2024-01-16 10:09:32,015 INFO [decode.py:564] batch 160/?, cuts processed until now is 5875
2024-01-16 10:09:44,614 INFO [decode.py:564] batch 180/?, cuts processed until now is 6581
2024-01-16 10:09:56,882 INFO [decode.py:564] batch 200/?, cuts processed until now is 7122
2024-01-16 10:09:58,670 INFO [decode.py:580] The transcripts are stored in zipformer_bbpe/exp-context-size-2-lr-epochs-10-spec-aug-20-disable-musan/fast_beam_search/recogs-test-blank_penalty_0.0_beam_20.0_max_contexts_8_max_states_64-epoch-40-avg-10-beam-20.0-max-contexts-8-max-states-64-blank-penalty-0.0-use-averaged-model.txt
2024-01-16 10:09:58,892 INFO [utils.py:565] [test-blank_penalty_0.0_beam_20.0_max_contexts_8_max_states_64] %WER 4.43% [4646 / 104765, 81 ins, 337 del, 4228 sub ]
2024-01-16 10:09:59,240 INFO [decode.py:598] Wrote detailed error stats to zipformer_bbpe/exp-context-size-2-lr-epochs-10-spec-aug-20-disable-musan/fast_beam_search/errs-test-blank_penalty_0.0_beam_20.0_max_contexts_8_max_states_64-epoch-40-avg-10-beam-20.0-max-contexts-8-max-states-64-blank-penalty-0.0-use-averaged-model.txt
2024-01-16 10:09:59,243 INFO [decode.py:614] 
For test, WER of different settings are:
blank_penalty_0.0_beam_20.0_max_contexts_8_max_states_64	4.43	best for test

2024-01-16 10:09:59,250 INFO [decode.py:832] Done!