2023-10-24 16:26:52,077 INFO [decode.py:664] Decoding started
2023-10-24 16:26:52,077 INFO [decode.py:670] Device: cuda:0
2023-10-24 16:26:52,082 INFO [decode.py:680] {'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': '1.24.3', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': '821ebc378e7fb99b8adc81950227963332821e01', 'k2-git-date': 'Wed Jul 19 15:38:25 2023', 'lhotse-version': '1.16.0.dev+git.1db4d97a.clean', 'torch-version': '1.11.0+cu102', 'torch-cuda-available': True, 'torch-cuda-version': '10.2', 'python-version': '3.9', 'icefall-git-branch': 'dev_zipformer_cn', 'icefall-git-sha1': '6eb141f0-clean', 'icefall-git-date': 'Tue Oct 24 11:01:44 2023', '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/k2-1.24.3.dev20230721+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-1.16.0.dev0+git.1db4d97a.clean-py3.9.egg/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-2-0423201334-6587bbc68d-tn554', 'IP address': '10.177.74.211'}, 'epoch': 20, 'iter': 0, 'avg': 1, 'use_averaged_model': True, 'exp_dir': PosixPath('zipformer/exp-w-ctc'), 'bpe_model': 'data/lang_bpe_2000/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_2000'), 'decoding_method': 'greedy_search', 'beam_size': 4, 'beam': 20.0, 'ngram_lm_scale': 0.01, 'max_contexts': 8, 'max_states': 64, 'context_size': 2, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, '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', 'use_transducer': True, 'use_ctc': True, 'manifest_dir': PosixPath('data/fbank'), 'max_duration': 300.0, '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, 'input_strategy': 'PrecomputedFeatures', 'res_dir': PosixPath('zipformer/exp-w-ctc/greedy_search'), 'suffix': 'epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 2000}
2023-10-24 16:26:52,082 INFO [decode.py:682] About to create model
2023-10-24 16:26:52,656 INFO [decode.py:749] Calculating the averaged model over epoch range from 19 (excluded) to 20
2023-10-24 16:27:00,102 INFO [decode.py:783] Number of model parameters: 69651511
2023-10-24 16:27:00,102 INFO [multi_dataset.py:221] About to get multidataset test cuts
2023-10-24 16:27:00,102 INFO [multi_dataset.py:224] Loading Aidatatang_200zh set in lazy mode
2023-10-24 16:27:00,196 INFO [multi_dataset.py:233] Loading Aishell set in lazy mode
2023-10-24 16:27:00,249 INFO [multi_dataset.py:242] Loading Aishell-2 set in lazy mode
2023-10-24 16:27:00,291 INFO [multi_dataset.py:251] Loading Aishell-4 TEST set in lazy mode
2023-10-24 16:27:00,317 INFO [multi_dataset.py:257] Loading Ali-Meeting set in lazy mode
2023-10-24 16:27:00,356 INFO [multi_dataset.py:266] Loading MagicData set in lazy mode
2023-10-24 16:27:00,422 INFO [multi_dataset.py:275] Loading KeSpeech set in lazy mode
2023-10-24 16:27:00,499 INFO [multi_dataset.py:287] Loading WeNetSpeech set in lazy mode
2023-10-24 16:27:06,618 WARNING [decode.py:793] Excluding cut with ID: TEST_NET_Y0000000004_0ub4ZzdHzBc_S00023 from decoding, num_frames: 8
2023-10-24 16:27:07,562 INFO [decode.py:807] Start decoding test set: aidatatang_test
2023-10-24 16:27:09,222 INFO [decode.py:561] batch 0/?, cuts processed until now is 80
2023-10-24 16:27:24,182 INFO [decode.py:561] batch 50/?, cuts processed until now is 4543
2023-10-24 16:27:39,845 INFO [decode.py:561] batch 100/?, cuts processed until now is 9084
2023-10-24 16:27:46,355 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([3.4148, 2.7863, 1.6394, 2.0368], device='cuda:0')
2023-10-24 16:27:47,605 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([3.1115, 3.1993, 2.0795, 2.0810], device='cuda:0')
2023-10-24 16:27:54,909 INFO [decode.py:561] batch 150/?, cuts processed until now is 13880
2023-10-24 16:28:10,246 INFO [decode.py:561] batch 200/?, cuts processed until now is 18516
2023-10-24 16:28:12,321 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.4605, 3.1105, 3.4316, 3.0579], device='cuda:0')
2023-10-24 16:28:18,978 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([4.5857, 4.4230, 4.0916, 4.7478], device='cuda:0')
2023-10-24 16:28:25,170 INFO [decode.py:561] batch 250/?, cuts processed until now is 22994
2023-10-24 16:28:39,936 INFO [decode.py:561] batch 300/?, cuts processed until now is 28179
2023-10-24 16:28:46,579 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([1.6580, 2.1367, 2.2847, 2.0372, 2.2609, 2.3253, 1.9475, 2.1614],
       device='cuda:0')
2023-10-24 16:28:54,691 INFO [decode.py:561] batch 350/?, cuts processed until now is 32886
2023-10-24 16:29:08,978 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.1169, 2.4798, 2.8846, 2.4491], device='cuda:0')
2023-10-24 16:29:09,826 INFO [decode.py:561] batch 400/?, cuts processed until now is 37667
2023-10-24 16:29:24,688 INFO [decode.py:561] batch 450/?, cuts processed until now is 42122
2023-10-24 16:29:39,766 INFO [decode.py:561] batch 500/?, cuts processed until now is 46172
2023-10-24 16:29:52,280 INFO [decode.py:577] The transcripts are stored in zipformer/exp-w-ctc/greedy_search/recogs-aidatatang_test-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:29:53,235 INFO [utils.py:565] [aidatatang_test-greedy_search] %WER 3.83% [17980 / 468933, 734 ins, 5370 del, 11876 sub ]
2023-10-24 16:29:54,953 INFO [decode.py:590] Wrote detailed error stats to zipformer/exp-w-ctc/greedy_search/errs-aidatatang_test-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:29:54,957 INFO [decode.py:606] 
For aidatatang_test, WER of different settings are:
greedy_search	3.83	best for aidatatang_test

2023-10-24 16:29:54,958 INFO [decode.py:807] Start decoding test set: aidatatang_dev
2023-10-24 16:29:56,879 INFO [decode.py:561] batch 0/?, cuts processed until now is 81
2023-10-24 16:30:11,861 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([1.6306, 2.1290, 2.2852, 2.0017, 2.2532, 2.3424, 1.9270, 2.1533],
       device='cuda:0')
2023-10-24 16:30:12,032 INFO [decode.py:561] batch 50/?, cuts processed until now is 4556
2023-10-24 16:30:16,150 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([4.0290, 2.3391, 4.0876, 3.5964], device='cuda:0')
2023-10-24 16:30:27,070 INFO [decode.py:561] batch 100/?, cuts processed until now is 9077
2023-10-24 16:30:42,120 INFO [decode.py:561] batch 150/?, cuts processed until now is 13773
2023-10-24 16:30:56,454 INFO [decode.py:561] batch 200/?, cuts processed until now is 18432
2023-10-24 16:31:02,823 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([1.7963, 2.4572, 2.6171, 2.5090, 2.3973, 2.3723, 1.6846, 2.4017],
       device='cuda:0')
2023-10-24 16:31:11,523 INFO [decode.py:561] batch 250/?, cuts processed until now is 22590
2023-10-24 16:31:17,107 INFO [decode.py:577] The transcripts are stored in zipformer/exp-w-ctc/greedy_search/recogs-aidatatang_dev-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:31:17,564 INFO [utils.py:565] [aidatatang_dev-greedy_search] %WER 3.36% [7875 / 234524, 309 ins, 2547 del, 5019 sub ]
2023-10-24 16:31:18,420 INFO [decode.py:590] Wrote detailed error stats to zipformer/exp-w-ctc/greedy_search/errs-aidatatang_dev-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:31:18,423 INFO [decode.py:606] 
For aidatatang_dev, WER of different settings are:
greedy_search	3.36	best for aidatatang_dev

2023-10-24 16:31:18,424 INFO [decode.py:807] Start decoding test set: alimeeting_test
2023-10-24 16:31:20,464 INFO [decode.py:561] batch 0/?, cuts processed until now is 44
2023-10-24 16:31:27,421 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.7918, 2.8049, 1.7817, 1.8504], device='cuda:0')
2023-10-24 16:31:39,166 WARNING [decode.py:793] Excluding cut with ID: R8008_M8016-8062-123 from decoding, num_frames: 6
2023-10-24 16:31:43,088 INFO [decode.py:561] batch 50/?, cuts processed until now is 3819
2023-10-24 16:31:46,527 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([1.1988, 1.8796, 2.0468, 1.7923, 1.8477, 2.0008, 1.7866, 1.8257],
       device='cuda:0')
2023-10-24 16:32:04,780 INFO [decode.py:561] batch 100/?, cuts processed until now is 7625
2023-10-24 16:32:24,790 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([4.6583, 4.5502, 4.3398, 4.6992], device='cuda:0')
2023-10-24 16:32:27,211 INFO [decode.py:561] batch 150/?, cuts processed until now is 11772
2023-10-24 16:32:38,627 INFO [decode.py:577] The transcripts are stored in zipformer/exp-w-ctc/greedy_search/recogs-alimeeting_test-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:32:39,176 INFO [utils.py:565] [alimeeting_test-greedy_search] %WER 25.18% [52829 / 209845, 4316 ins, 26624 del, 21889 sub ]
2023-10-24 16:32:40,166 INFO [decode.py:590] Wrote detailed error stats to zipformer/exp-w-ctc/greedy_search/errs-alimeeting_test-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:32:40,170 INFO [decode.py:606] 
For alimeeting_test, WER of different settings are:
greedy_search	25.18	best for alimeeting_test

2023-10-24 16:32:40,170 INFO [decode.py:807] Start decoding test set: alimeeting_eval
2023-10-24 16:32:41,957 INFO [decode.py:561] batch 0/?, cuts processed until now is 35
2023-10-24 16:32:48,932 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([1.4953, 3.3299, 3.0715, 3.1702], device='cuda:0')
2023-10-24 16:33:05,650 INFO [decode.py:561] batch 50/?, cuts processed until now is 3467
2023-10-24 16:33:12,792 INFO [decode.py:577] The transcripts are stored in zipformer/exp-w-ctc/greedy_search/recogs-alimeeting_eval-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:33:13,009 INFO [utils.py:565] [alimeeting_eval-greedy_search] %WER 23.90% [19385 / 81111, 1756 ins, 9271 del, 8358 sub ]
2023-10-24 16:33:13,335 INFO [decode.py:590] Wrote detailed error stats to zipformer/exp-w-ctc/greedy_search/errs-alimeeting_eval-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:33:13,339 INFO [decode.py:606] 
For alimeeting_eval, WER of different settings are:
greedy_search	23.9	best for alimeeting_eval

2023-10-24 16:33:13,340 INFO [decode.py:807] Start decoding test set: aishell_test
2023-10-24 16:33:14,892 INFO [decode.py:561] batch 0/?, cuts processed until now is 47
2023-10-24 16:33:33,791 INFO [decode.py:561] batch 50/?, cuts processed until now is 2712
2023-10-24 16:33:52,165 INFO [decode.py:561] batch 100/?, cuts processed until now is 5468
2023-10-24 16:34:04,606 INFO [decode.py:577] The transcripts are stored in zipformer/exp-w-ctc/greedy_search/recogs-aishell_test-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:34:04,789 INFO [utils.py:565] [aishell_test-greedy_search] %WER 3.08% [3223 / 104765, 76 ins, 1157 del, 1990 sub ]
2023-10-24 16:34:05,154 INFO [decode.py:590] Wrote detailed error stats to zipformer/exp-w-ctc/greedy_search/errs-aishell_test-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:34:05,158 INFO [decode.py:606] 
For aishell_test, WER of different settings are:
greedy_search	3.08	best for aishell_test

2023-10-24 16:34:05,158 INFO [decode.py:807] Start decoding test set: aishell_dev
2023-10-24 16:34:06,806 INFO [decode.py:561] batch 0/?, cuts processed until now is 53
2023-10-24 16:34:10,612 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([1.7319, 1.9604, 2.2042, 1.0559], device='cuda:0')
2023-10-24 16:34:24,880 INFO [decode.py:561] batch 50/?, cuts processed until now is 3030
2023-10-24 16:34:37,771 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([1.8683, 2.7552, 2.8765, 2.6969, 2.7064, 2.7163, 1.6722, 2.6946],
       device='cuda:0')
2023-10-24 16:34:39,949 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([3.4992, 3.4650, 2.5676, 2.4659], device='cuda:0')
2023-10-24 16:34:42,875 INFO [decode.py:561] batch 100/?, cuts processed until now is 6034
2023-10-24 16:34:59,799 INFO [decode.py:561] batch 150/?, cuts processed until now is 9180
2023-10-24 16:35:05,735 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.6986, 3.0384, 3.5315, 2.2964], device='cuda:0')
2023-10-24 16:35:17,283 INFO [decode.py:561] batch 200/?, cuts processed until now is 12198
2023-10-24 16:35:26,793 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.5987, 3.1012, 3.4631, 3.1286, 3.2995, 2.9732, 2.8481, 3.3857],
       device='cuda:0')
2023-10-24 16:35:30,886 INFO [decode.py:577] The transcripts are stored in zipformer/exp-w-ctc/greedy_search/recogs-aishell_dev-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:35:31,253 INFO [utils.py:565] [aishell_dev-greedy_search] %WER 2.77% [5695 / 205341, 193 ins, 1959 del, 3543 sub ]
2023-10-24 16:35:31,972 INFO [decode.py:590] Wrote detailed error stats to zipformer/exp-w-ctc/greedy_search/errs-aishell_dev-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:35:31,976 INFO [decode.py:606] 
For aishell_dev, WER of different settings are:
greedy_search	2.77	best for aishell_dev

2023-10-24 16:35:31,977 INFO [decode.py:807] Start decoding test set: aishell-2_test
2023-10-24 16:35:33,498 INFO [decode.py:561] batch 0/?, cuts processed until now is 83
2023-10-24 16:35:48,195 INFO [decode.py:561] batch 50/?, cuts processed until now is 4841
2023-10-24 16:35:48,242 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([1.7917, 2.0368, 1.9793, 3.2922], device='cuda:0')
2023-10-24 16:35:49,248 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([1.8701, 2.6559, 2.6194, 2.7435], device='cuda:0')
2023-10-24 16:35:49,704 INFO [decode.py:577] The transcripts are stored in zipformer/exp-w-ctc/greedy_search/recogs-aishell-2_test-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:35:49,803 INFO [utils.py:565] [aishell-2_test-greedy_search] %WER 4.04% [2001 / 49532, 57 ins, 378 del, 1566 sub ]
2023-10-24 16:35:49,992 INFO [decode.py:590] Wrote detailed error stats to zipformer/exp-w-ctc/greedy_search/errs-aishell-2_test-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:35:49,995 INFO [decode.py:606] 
For aishell-2_test, WER of different settings are:
greedy_search	4.04	best for aishell-2_test

2023-10-24 16:35:49,996 INFO [decode.py:807] Start decoding test set: aishell-2_dev
2023-10-24 16:35:51,177 INFO [decode.py:561] batch 0/?, cuts processed until now is 81
2023-10-24 16:35:59,497 INFO [decode.py:577] The transcripts are stored in zipformer/exp-w-ctc/greedy_search/recogs-aishell-2_dev-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:35:59,566 INFO [utils.py:565] [aishell-2_dev-greedy_search] %WER 3.70% [918 / 24802, 14 ins, 166 del, 738 sub ]
2023-10-24 16:35:59,677 INFO [decode.py:590] Wrote detailed error stats to zipformer/exp-w-ctc/greedy_search/errs-aishell-2_dev-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:35:59,680 INFO [decode.py:606] 
For aishell-2_dev, WER of different settings are:
greedy_search	3.7	best for aishell-2_dev

2023-10-24 16:35:59,680 INFO [decode.py:807] Start decoding test set: aishell-4
2023-10-24 16:36:01,892 INFO [decode.py:561] batch 0/?, cuts processed until now is 33
2023-10-24 16:36:26,178 INFO [decode.py:561] batch 50/?, cuts processed until now is 2510
2023-10-24 16:36:37,771 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.8036, 2.7350, 2.6902, 5.0775], device='cuda:0')
2023-10-24 16:36:45,540 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.5539, 3.3920, 3.6542, 2.9110, 3.2302, 3.4420, 3.0393, 2.9827],
       device='cuda:0')
2023-10-24 16:36:47,359 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([1.8455, 2.0135, 1.8856, 2.0848], device='cuda:0')
2023-10-24 16:36:50,033 INFO [decode.py:561] batch 100/?, cuts processed until now is 5000
2023-10-24 16:36:53,727 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([4.9601, 2.8495, 4.9928, 4.5408], device='cuda:0')
2023-10-24 16:37:10,804 INFO [decode.py:561] batch 150/?, cuts processed until now is 8249
2023-10-24 16:37:21,014 INFO [decode.py:577] The transcripts are stored in zipformer/exp-w-ctc/greedy_search/recogs-aishell-4-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:37:21,342 INFO [utils.py:565] [aishell-4-greedy_search] %WER 16.13% [29144 / 180665, 4508 ins, 10175 del, 14461 sub ]
2023-10-24 16:37:22,011 INFO [decode.py:590] Wrote detailed error stats to zipformer/exp-w-ctc/greedy_search/errs-aishell-4-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:37:22,015 INFO [decode.py:606] 
For aishell-4, WER of different settings are:
greedy_search	16.13	best for aishell-4

2023-10-24 16:37:22,015 INFO [decode.py:807] Start decoding test set: magicdata_test
2023-10-24 16:37:23,638 INFO [decode.py:561] batch 0/?, cuts processed until now is 57
2023-10-24 16:37:40,572 INFO [decode.py:561] batch 50/?, cuts processed until now is 3245
2023-10-24 16:37:56,805 INFO [decode.py:561] batch 100/?, cuts processed until now is 6425
2023-10-24 16:38:08,793 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([1.9764, 2.5280, 2.4107, 2.3617, 2.5035, 2.5184, 2.2359, 2.3483],
       device='cuda:0')
2023-10-24 16:38:12,930 INFO [decode.py:561] batch 150/?, cuts processed until now is 9787
2023-10-24 16:38:29,072 INFO [decode.py:561] batch 200/?, cuts processed until now is 13211
2023-10-24 16:38:42,035 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.1606, 1.9393, 2.1190, 2.2631, 2.7297, 2.5162, 2.3656, 2.0564],
       device='cuda:0')
2023-10-24 16:38:45,293 INFO [decode.py:561] batch 250/?, cuts processed until now is 16395
2023-10-24 16:38:59,930 INFO [decode.py:561] batch 300/?, cuts processed until now is 20136
2023-10-24 16:39:00,950 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.8128, 4.0924, 4.0309, 4.0854], device='cuda:0')
2023-10-24 16:39:16,186 INFO [decode.py:561] batch 350/?, cuts processed until now is 23001
2023-10-24 16:39:22,317 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.5293, 3.0856, 3.1968, 2.9030, 3.2124, 2.7817, 2.3875, 2.7488],
       device='cuda:0')
2023-10-24 16:39:26,968 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.6346, 2.9238, 2.5255, 4.5718], device='cuda:0')
2023-10-24 16:39:27,359 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([5.2807, 4.4323, 5.3561, 4.9194], device='cuda:0')
2023-10-24 16:39:30,020 INFO [decode.py:577] The transcripts are stored in zipformer/exp-w-ctc/greedy_search/recogs-magicdata_test-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:39:30,497 INFO [utils.py:565] [magicdata_test-greedy_search] %WER 3.15% [7543 / 239091, 291 ins, 2093 del, 5159 sub ]
2023-10-24 16:39:31,461 INFO [decode.py:590] Wrote detailed error stats to zipformer/exp-w-ctc/greedy_search/errs-magicdata_test-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:39:31,465 INFO [decode.py:606] 
For magicdata_test, WER of different settings are:
greedy_search	3.15	best for magicdata_test

2023-10-24 16:39:31,466 INFO [decode.py:807] Start decoding test set: magicdata_dev
2023-10-24 16:39:33,149 INFO [decode.py:561] batch 0/?, cuts processed until now is 52
2023-10-24 16:39:49,865 INFO [decode.py:561] batch 50/?, cuts processed until now is 2982
2023-10-24 16:39:54,314 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.7383, 2.4002, 2.8558, 3.0812, 3.3021, 3.1870, 3.0472, 2.4360],
       device='cuda:0')
2023-10-24 16:40:06,350 INFO [decode.py:561] batch 100/?, cuts processed until now is 5919
2023-10-24 16:40:09,991 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.2431, 3.1154, 3.0753, 3.1581], device='cuda:0')
2023-10-24 16:40:10,595 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([3.1981, 3.5185, 2.5220, 4.7182], device='cuda:0')
2023-10-24 16:40:19,854 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.5817, 3.0980, 3.5288, 3.7128], device='cuda:0')
2023-10-24 16:40:22,959 INFO [decode.py:561] batch 150/?, cuts processed until now is 9079
2023-10-24 16:40:30,657 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.4665, 3.5307, 3.5353, 3.5293], device='cuda:0')
2023-10-24 16:40:37,645 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.3204, 2.7968, 3.2415, 3.4578], device='cuda:0')
2023-10-24 16:40:38,517 INFO [decode.py:561] batch 200/?, cuts processed until now is 11646
2023-10-24 16:40:40,191 INFO [decode.py:577] The transcripts are stored in zipformer/exp-w-ctc/greedy_search/recogs-magicdata_dev-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:40:40,418 INFO [utils.py:565] [magicdata_dev-greedy_search] %WER 3.77% [4407 / 116800, 251 ins, 1085 del, 3071 sub ]
2023-10-24 16:40:40,861 INFO [decode.py:590] Wrote detailed error stats to zipformer/exp-w-ctc/greedy_search/errs-magicdata_dev-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:40:40,865 INFO [decode.py:606] 
For magicdata_dev, WER of different settings are:
greedy_search	3.77	best for magicdata_dev

2023-10-24 16:40:40,865 INFO [decode.py:807] Start decoding test set: kespeech-asr_test
2023-10-24 16:40:42,760 INFO [decode.py:561] batch 0/?, cuts processed until now is 45
2023-10-24 16:40:47,680 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.1763, 2.7182, 2.9618, 2.7043, 2.8261, 2.6437, 2.9456, 2.8986],
       device='cuda:0')
2023-10-24 16:40:56,484 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.5361, 3.8369, 3.8190, 3.6853], device='cuda:0')
2023-10-24 16:41:02,081 INFO [decode.py:561] batch 50/?, cuts processed until now is 2446
2023-10-24 16:41:19,981 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.9551, 4.4044, 3.6287, 3.9220], device='cuda:0')
2023-10-24 16:41:21,528 INFO [decode.py:561] batch 100/?, cuts processed until now is 4867
2023-10-24 16:41:22,828 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([5.0844, 3.6308, 5.3723, 4.8003], device='cuda:0')
2023-10-24 16:41:40,478 INFO [decode.py:561] batch 150/?, cuts processed until now is 7421
2023-10-24 16:41:59,384 INFO [decode.py:561] batch 200/?, cuts processed until now is 9965
2023-10-24 16:42:18,338 INFO [decode.py:561] batch 250/?, cuts processed until now is 12394
2023-10-24 16:42:36,414 INFO [decode.py:561] batch 300/?, cuts processed until now is 15124
2023-10-24 16:42:55,369 INFO [decode.py:561] batch 350/?, cuts processed until now is 17516
2023-10-24 16:43:10,825 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([5.6071, 5.1288, 5.4883, 5.1307], device='cuda:0')
2023-10-24 16:43:12,542 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([5.9149, 5.8442, 5.5428, 6.1460], device='cuda:0')
2023-10-24 16:43:13,800 INFO [decode.py:561] batch 400/?, cuts processed until now is 19643
2023-10-24 16:43:14,960 INFO [decode.py:577] The transcripts are stored in zipformer/exp-w-ctc/greedy_search/recogs-kespeech-asr_test-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:43:15,523 INFO [utils.py:565] [kespeech-asr_test-greedy_search] %WER 8.08% [22928 / 283772, 1266 ins, 3566 del, 18096 sub ]
2023-10-24 16:43:16,552 INFO [decode.py:590] Wrote detailed error stats to zipformer/exp-w-ctc/greedy_search/errs-kespeech-asr_test-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:43:16,557 INFO [decode.py:606] 
For kespeech-asr_test, WER of different settings are:
greedy_search	8.08	best for kespeech-asr_test

2023-10-24 16:43:16,557 INFO [decode.py:807] Start decoding test set: kespeech-asr_dev_phase1
2023-10-24 16:43:17,795 INFO [decode.py:561] batch 0/?, cuts processed until now is 44
2023-10-24 16:43:23,294 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([3.6919, 3.7749, 2.9059, 2.7458], device='cuda:0')
2023-10-24 16:43:32,522 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.0980, 2.7457, 3.2170, 3.1968], device='cuda:0')
2023-10-24 16:43:34,798 INFO [decode.py:577] The transcripts are stored in zipformer/exp-w-ctc/greedy_search/recogs-kespeech-asr_dev_phase1-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:43:34,861 INFO [utils.py:565] [kespeech-asr_dev_phase1-greedy_search] %WER 6.88% [2176 / 31634, 133 ins, 375 del, 1668 sub ]
2023-10-24 16:43:34,979 INFO [decode.py:590] Wrote detailed error stats to zipformer/exp-w-ctc/greedy_search/errs-kespeech-asr_dev_phase1-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:43:34,983 INFO [decode.py:606] 
For kespeech-asr_dev_phase1, WER of different settings are:
greedy_search	6.88	best for kespeech-asr_dev_phase1

2023-10-24 16:43:34,983 INFO [decode.py:807] Start decoding test set: kespeech-asr_dev_phase2
2023-10-24 16:43:36,155 INFO [decode.py:561] batch 0/?, cuts processed until now is 47
2023-10-24 16:43:53,125 INFO [decode.py:577] The transcripts are stored in zipformer/exp-w-ctc/greedy_search/recogs-kespeech-asr_dev_phase2-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:43:53,185 INFO [utils.py:565] [kespeech-asr_dev_phase2-greedy_search] %WER 3.14% [1002 / 31928, 53 ins, 224 del, 725 sub ]
2023-10-24 16:43:53,301 INFO [decode.py:590] Wrote detailed error stats to zipformer/exp-w-ctc/greedy_search/errs-kespeech-asr_dev_phase2-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:43:53,305 INFO [decode.py:606] 
For kespeech-asr_dev_phase2, WER of different settings are:
greedy_search	3.14	best for kespeech-asr_dev_phase2

2023-10-24 16:43:53,305 INFO [decode.py:807] Start decoding test set: wenetspeech-meeting_test
2023-10-24 16:43:55,222 INFO [decode.py:561] batch 0/?, cuts processed until now is 28
2023-10-24 16:44:20,978 INFO [decode.py:561] batch 50/?, cuts processed until now is 1884
2023-10-24 16:44:24,375 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.6509, 3.5233, 3.6944, 2.9665, 3.4160, 3.4758, 3.1022, 3.1446],
       device='cuda:0')
2023-10-24 16:44:30,130 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([1.7567, 3.4571, 3.2859, 4.6832], device='cuda:0')
2023-10-24 16:44:47,585 INFO [decode.py:561] batch 100/?, cuts processed until now is 3776
2023-10-24 16:45:06,756 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([1.8404, 2.4345, 2.5365, 2.4522, 2.2006, 2.3045, 1.7031, 2.0973],
       device='cuda:0')
2023-10-24 16:45:12,919 INFO [decode.py:561] batch 150/?, cuts processed until now is 5887
2023-10-24 16:45:37,667 INFO [decode.py:561] batch 200/?, cuts processed until now is 8092
2023-10-24 16:45:41,847 INFO [decode.py:577] The transcripts are stored in zipformer/exp-w-ctc/greedy_search/recogs-wenetspeech-meeting_test-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:45:42,213 INFO [utils.py:565] [wenetspeech-meeting_test-greedy_search] %WER 7.19% [15845 / 220385, 1568 ins, 5007 del, 9270 sub ]
2023-10-24 16:45:42,939 INFO [decode.py:590] Wrote detailed error stats to zipformer/exp-w-ctc/greedy_search/errs-wenetspeech-meeting_test-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:45:42,942 INFO [decode.py:606] 
For wenetspeech-meeting_test, WER of different settings are:
greedy_search	7.19	best for wenetspeech-meeting_test

2023-10-24 16:45:42,943 INFO [decode.py:807] Start decoding test set: wenetspeech-net_test
2023-10-24 16:45:43,202 WARNING [decode.py:793] Excluding cut with ID: TEST_NET_Y0000000004_0ub4ZzdHzBc_S00023 from decoding, num_frames: 8
2023-10-24 16:45:44,934 INFO [decode.py:561] batch 0/?, cuts processed until now is 43
2023-10-24 16:46:04,905 INFO [decode.py:561] batch 50/?, cuts processed until now is 3497
2023-10-24 16:46:10,418 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.2307, 3.1409, 3.5551, 3.2214, 3.2725, 2.7590, 2.1443, 2.9630],
       device='cuda:0')
2023-10-24 16:46:23,679 INFO [decode.py:561] batch 100/?, cuts processed until now is 7009
2023-10-24 16:46:32,546 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.9441, 2.9995, 1.9044, 1.9497], device='cuda:0')
2023-10-24 16:46:43,198 INFO [decode.py:561] batch 150/?, cuts processed until now is 11030
2023-10-24 16:46:49,567 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([1.1613, 1.6426, 1.7442, 2.2402], device='cuda:0')
2023-10-24 16:47:02,612 INFO [decode.py:561] batch 200/?, cuts processed until now is 14995
2023-10-24 16:47:22,547 INFO [decode.py:561] batch 250/?, cuts processed until now is 18681
2023-10-24 16:47:39,725 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([1.6539, 2.5277, 2.7432, 2.1305, 2.5100, 2.6285, 2.1877, 2.3297],
       device='cuda:0')
2023-10-24 16:47:41,282 INFO [decode.py:561] batch 300/?, cuts processed until now is 22693
2023-10-24 16:47:50,840 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([0.9792, 1.6031, 1.6654, 1.4956, 1.6311, 1.6562, 1.5561, 1.4501],
       device='cuda:0')
2023-10-24 16:47:53,063 INFO [decode.py:577] The transcripts are stored in zipformer/exp-w-ctc/greedy_search/recogs-wenetspeech-net_test-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:47:53,809 INFO [utils.py:565] [wenetspeech-net_test-greedy_search] %WER 8.17% [33964 / 415746, 1516 ins, 14317 del, 18131 sub ]
2023-10-24 16:47:55,365 INFO [decode.py:590] Wrote detailed error stats to zipformer/exp-w-ctc/greedy_search/errs-wenetspeech-net_test-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:47:55,370 INFO [decode.py:606] 
For wenetspeech-net_test, WER of different settings are:
greedy_search	8.17	best for wenetspeech-net_test

2023-10-24 16:47:55,370 INFO [decode.py:807] Start decoding test set: wenetspeech_dev
2023-10-24 16:47:57,289 INFO [decode.py:561] batch 0/?, cuts processed until now is 39
2023-10-24 16:48:20,013 INFO [decode.py:561] batch 50/?, cuts processed until now is 2499
2023-10-24 16:48:34,470 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([2.1800, 2.8665, 3.4702, 3.4613], device='cuda:0')
2023-10-24 16:48:38,177 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([1.9991, 2.3058, 2.7628, 2.3184, 2.3961, 2.1567, 2.5103, 2.5484],
       device='cuda:0')
2023-10-24 16:48:41,947 INFO [decode.py:561] batch 100/?, cuts processed until now is 4983
2023-10-24 16:49:02,897 INFO [decode.py:561] batch 150/?, cuts processed until now is 7682
2023-10-24 16:49:24,329 INFO [decode.py:561] batch 200/?, cuts processed until now is 10268
2023-10-24 16:49:44,821 INFO [decode.py:561] batch 250/?, cuts processed until now is 13004
2023-10-24 16:49:49,994 INFO [decode.py:577] The transcripts are stored in zipformer/exp-w-ctc/greedy_search/recogs-wenetspeech_dev-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:49:50,547 INFO [utils.py:565] [wenetspeech_dev-greedy_search] %WER 9.04% [29889 / 330498, 1369 ins, 18677 del, 9843 sub ]
2023-10-24 16:49:51,646 INFO [decode.py:590] Wrote detailed error stats to zipformer/exp-w-ctc/greedy_search/errs-wenetspeech_dev-greedy_search-epoch-20-avg-1-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-10-24 16:49:51,649 INFO [decode.py:606] 
For wenetspeech_dev, WER of different settings are:
greedy_search	9.04	best for wenetspeech_dev

2023-10-24 16:49:51,650 INFO [decode.py:824] Done!