decoding-results/greedy_search/log-decode-epoch-30-avg-11-context-2-max-sym-per-frame-1-use-averaged-model-2023-03-10-10-18-49

2023-03-10 10:18:49,578 INFO [decode_with_timestamp.py:878] Decoding started
2023-03-10 10:18:49,578 INFO [decode_with_timestamp.py:884] Device: cuda:0
2023-03-10 10:18:49,581 INFO [decode_with_timestamp.py:899] {'frame_shift_ms': 10.0, 'allowed_excess_duration_ratio': 0.1, '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': '1.22', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': '96c9a2aece2a3a7633da07740e24fa3d96f5498c', 'k2-git-date': 'Thu Nov 10 08:14:02 2022', 'lhotse-version': '1.13.0.dev+git.527d964.clean', 'torch-version': '1.12.1', 'torch-cuda-available': True, 'torch-cuda-version': '11.6', 'python-version': '3.8', 'icefall-git-branch': 'random_padding', 'icefall-git-sha1': '202ce08-clean', 'icefall-git-date': 'Thu Mar 9 15:05:03 2023', 'icefall-path': '/ceph-data4/yangxiaoyu/softwares/icefall_development/icefall_random_padding', 'k2-path': '/ceph-data4/yangxiaoyu/softwares/anaconda3/envs/k2_latest/lib/python3.8/site-packages/k2/__init__.py', 'lhotse-path': '/ceph-data4/yangxiaoyu/softwares/lhotse_development/lhotse_random_padding_left/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-2-1216192652-5bcf7587b4-n6q9m', 'IP address': '10.177.74.211'}, 'epoch': 30, 'iter': 0, 'avg': 11, 'use_averaged_model': True, 'exp_dir': PosixPath('pruned_transducer_stateless7/exp_960h_no_paddingidx_ngpu4'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), '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, 'simulate_streaming': False, 'decode_chunk_size': 16, 'left_context': 64, 'use_shallow_fusion': False, 'lm_type': 'rnn', 'lm_scale': 0.3, 'tokens_ngram': 3, 'backoff_id': 500, 'num_encoder_layers': '2,4,3,2,4', 'feedforward_dims': '1024,1024,2048,2048,1024', 'nhead': '8,8,8,8,8', 'encoder_dims': '384,384,384,384,384', 'attention_dims': '192,192,192,192,192', 'encoder_unmasked_dims': '256,256,256,256,256', 'zipformer_downsampling_factors': '1,2,4,8,2', 'cnn_module_kernels': '31,31,31,31,31', 'decoder_dim': 512, 'joiner_dim': 512, 'full_libri': True, 'manifest_dir': PosixPath('data/fbank_ali'), 'max_duration': 500, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'random_left_padding': False, 'num_left_padding': 8, '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', 'vocab_size': 500, 'lm_epoch': 7, 'lm_avg': 1, 'lm_exp_dir': None, 'rnn_lm_embedding_dim': 2048, 'rnn_lm_hidden_dim': 2048, 'rnn_lm_num_layers': 3, 'rnn_lm_tie_weights': True, 'transformer_lm_exp_dir': None, 'transformer_lm_dim_feedforward': 2048, 'transformer_lm_encoder_dim': 768, 'transformer_lm_embedding_dim': 768, 'transformer_lm_nhead': 8, 'transformer_lm_num_layers': 16, 'transformer_lm_tie_weights': True, 'res_dir': PosixPath('pruned_transducer_stateless7/exp_960h_no_paddingidx_ngpu4/greedy_search'), 'suffix': 'epoch-30-avg-11-context-2-max-sym-per-frame-1-use-averaged-model', 'blank_id': 0, 'unk_id': 2}
2023-03-10 10:18:49,581 INFO [decode_with_timestamp.py:901] About to create model
2023-03-10 10:18:50,174 INFO [zipformer.py:178] At encoder stack 4, which has downsampling_factor=2, we will combine the outputs of layers 1 and 3, with downsampling_factors=2 and 8.
2023-03-10 10:18:50,187 INFO [decode_with_timestamp.py:968] Calculating the averaged model over epoch range from 19 (excluded) to 30
2023-03-10 10:18:59,777 INFO [decode_with_timestamp.py:1030] Number of model parameters: 70369391
2023-03-10 10:18:59,778 INFO [asr_datamodule.py:463] About to get test-clean cuts
2023-03-10 10:18:59,781 INFO [asr_datamodule.py:470] About to get test-other cuts
2023-03-10 10:19:04,581 INFO [decode_with_timestamp.py:740] batch 0/?, cuts processed until now is 36
2023-03-10 10:20:02,828 INFO [decode_with_timestamp.py:740] batch 50/?, cuts processed until now is 2611
2023-03-10 10:20:04,125 INFO [decode_with_timestamp.py:1062] Averaged first symbol emission time: 0.09091603053435197
2023-03-10 10:20:04,203 INFO [decode_with_timestamp.py:760] The transcripts are stored in pruned_transducer_stateless7/exp_960h_no_paddingidx_ngpu4/greedy_search/recogs-test-clean-greedy_search-epoch-30-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-03-10 10:20:04,334 INFO [utils.py:795] [test-clean-greedy_search] %WER 2.25% [1182 / 52576, 126 ins, 103 del, 953 sub ]
2023-03-10 10:20:04,334 INFO [utils.py:800] [test-clean-greedy_search] %symbol-delay mean (s): -0.044, variance: 0.007 computed on 51520 correct words
2023-03-10 10:20:04,631 INFO [decode_with_timestamp.py:774] Wrote detailed error stats to pruned_transducer_stateless7/exp_960h_no_paddingidx_ngpu4/greedy_search/errs-test-clean-greedy_search-epoch-30-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-03-10 10:20:04,633 INFO [decode_with_timestamp.py:803] 
For test-clean, WER of different settings are:
greedy_search	2.25	best for test-clean

2023-03-10 10:20:04,633 INFO [decode_with_timestamp.py:810] 
For test-clean, symbol-delay of different settings are:
greedy_search	mean: -0.044s, variance: 0.007	best for test-clean

2023-03-10 10:20:06,540 INFO [decode_with_timestamp.py:740] batch 0/?, cuts processed until now is 43
2023-03-10 10:20:49,581 INFO [zipformer.py:1455] attn_weights_entropy = tensor([3.4986, 2.8964, 3.9553, 3.3571, 2.6642, 4.2705, 3.7907, 2.8587],
       device='cuda:0'), covar=tensor([0.0571, 0.1424, 0.0377, 0.0551, 0.1724, 0.0229, 0.0588, 0.0908],
       device='cuda:0'), in_proj_covar=tensor([0.0205, 0.0234, 0.0213, 0.0158, 0.0218, 0.0205, 0.0243, 0.0190],
       device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002],
       device='cuda:0')
2023-03-10 10:20:51,314 INFO [zipformer.py:1455] attn_weights_entropy = tensor([3.6831, 3.6830, 3.7112, 3.5030, 3.5668, 3.5269, 3.7213, 3.7422],
       device='cuda:0'), covar=tensor([0.0096, 0.0076, 0.0089, 0.0122, 0.0089, 0.0168, 0.0095, 0.0110],
       device='cuda:0'), in_proj_covar=tensor([0.0095, 0.0069, 0.0075, 0.0094, 0.0075, 0.0104, 0.0087, 0.0086],
       device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003],
       device='cuda:0')
2023-03-10 10:21:01,125 INFO [decode_with_timestamp.py:740] batch 50/?, cuts processed until now is 2939
2023-03-10 10:21:01,241 INFO [decode_with_timestamp.py:1062] Averaged first symbol emission time: 0.11329023477373397
2023-03-10 10:21:01,321 INFO [decode_with_timestamp.py:760] The transcripts are stored in pruned_transducer_stateless7/exp_960h_no_paddingidx_ngpu4/greedy_search/recogs-test-other-greedy_search-epoch-30-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-03-10 10:21:01,454 INFO [utils.py:795] [test-other-greedy_search] %WER 5.23% [2735 / 52343, 265 ins, 232 del, 2238 sub ]
2023-03-10 10:21:01,454 INFO [utils.py:800] [test-other-greedy_search] %symbol-delay mean (s): -0.05, variance: 0.008 computed on 49855 correct words
2023-03-10 10:21:01,666 INFO [decode_with_timestamp.py:774] Wrote detailed error stats to pruned_transducer_stateless7/exp_960h_no_paddingidx_ngpu4/greedy_search/errs-test-other-greedy_search-epoch-30-avg-11-context-2-max-sym-per-frame-1-use-averaged-model.txt
2023-03-10 10:21:01,667 INFO [decode_with_timestamp.py:803] 
For test-other, WER of different settings are:
greedy_search	5.23	best for test-other

2023-03-10 10:21:01,667 INFO [decode_with_timestamp.py:810] 
For test-other, symbol-delay of different settings are:
greedy_search	mean: -0.05s, variance: 0.008	best for test-other

2023-03-10 10:21:01,667 INFO [decode_with_timestamp.py:1071] Done!