2023-04-06 11:48:15,413 INFO [decode.py:659] Decoding started
2023-04-06 11:48:15,413 INFO [decode.py:665] Device: cuda:0
2023-04-06 11:48:15,420 INFO [decode.py:675] {'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': 'zipformer_libri_small_models', 'icefall-git-sha1': 'b52e7ae-dirty', 'icefall-git-date': 'Tue Apr 4 14:07:45 2023', 'icefall-path': '/ceph-data4/yangxiaoyu/softwares/icefall_development/icefall_small_models', '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-6-1219221738-65dd59bbf8-2ghmr', 'IP address': '10.177.28.85'}, 'epoch': 20, 'iter': 0, 'avg': 4, 'use_averaged_model': True, 'exp_dir': PosixPath('pruned_transducer_stateless7_streaming_multi/exp'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'modified_beam_search', 'beam_size': 4, 'beam': 20.0, 'ngram_lm_scale': 0.01, 'max_contexts': 8, 'max_states': 64, 'context_size': 2, 'right_padding': 64, 'max_sym_per_frame': 1, 'num_paths': 200, 'nbest_scale': 0.5, '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, 'short_chunk_size': 50, 'num_left_chunks': 4, 'decode_chunk_len': 32, 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'shuffle': True, 'return_cuts': True, 'num_workers': 2, 'on_the_fly_num_workers': 0, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'manifest_dir': PosixPath('data/fbank'), 'on_the_fly_feats': False, 'res_dir': PosixPath('pruned_transducer_stateless7_streaming_multi/exp/modified_beam_search'), 'suffix': 'epoch-20-avg-4-streaming-chunk-size-32-modified_beam_search-beam-size-4-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
2023-04-06 11:48:15,421 INFO [decode.py:677] About to create model
2023-04-06 11:48:16,250 INFO [zipformer.py:405] 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.
2023-04-06 11:48:16,269 INFO [decode.py:748] Calculating the averaged model over epoch range from 16 (excluded) to 20
2023-04-06 11:48:21,626 INFO [decode.py:782] Number of model parameters: 70369391
2023-04-06 11:48:21,626 INFO [librispeech.py:58] About to get test-clean cuts from data/fbank/librispeech_cuts_test-clean.jsonl.gz
2023-04-06 11:48:21,635 INFO [librispeech.py:63] About to get test-other cuts from data/fbank/librispeech_cuts_test-other.jsonl.gz
2023-04-06 11:48:33,261 INFO [decode.py:569] batch 0/?, cuts processed until now is 26
2023-04-06 11:50:53,265 INFO [decode.py:569] batch 20/?, cuts processed until now is 1545
2023-04-06 11:52:10,551 INFO [decode.py:569] batch 40/?, cuts processed until now is 2375
2023-04-06 11:52:51,714 INFO [decode.py:583] The transcripts are stored in pruned_transducer_stateless7_streaming_multi/exp/modified_beam_search/recogs-test-clean-epoch-20-avg-4-streaming-chunk-size-32-modified_beam_search-beam-size-4-use-averaged-model.txt
2023-04-06 11:52:51,860 INFO [utils.py:558] [test-clean-beam_size_4] %WER 2.40% [1262 / 52576, 151 ins, 95 del, 1016 sub ]
2023-04-06 11:52:52,205 INFO [decode.py:594] Wrote detailed error stats to pruned_transducer_stateless7_streaming_multi/exp/modified_beam_search/errs-test-clean-epoch-20-avg-4-streaming-chunk-size-32-modified_beam_search-beam-size-4-use-averaged-model.txt
2023-04-06 11:52:52,206 INFO [decode.py:608] 
For test-clean, WER of different settings are:
beam_size_4	2.4	best for test-clean

2023-04-06 11:53:00,716 INFO [decode.py:569] batch 0/?, cuts processed until now is 30
2023-04-06 11:55:14,199 INFO [decode.py:569] batch 20/?, cuts processed until now is 1771
2023-04-06 11:56:37,308 INFO [decode.py:569] batch 40/?, cuts processed until now is 2696
2023-04-06 11:57:06,781 INFO [zipformer.py:2441] attn_weights_entropy = tensor([3.5733, 3.4374, 2.7541, 3.9912, 3.4847, 3.5402, 1.5555, 3.4687],
       device='cuda:0'), covar=tensor([0.1611, 0.1036, 0.3460, 0.1373, 0.3198, 0.2362, 0.6204, 0.2127],
       device='cuda:0'), in_proj_covar=tensor([0.0265, 0.0238, 0.0290, 0.0333, 0.0331, 0.0273, 0.0293, 0.0287],
       device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002],
       device='cuda:0')
2023-04-06 11:57:15,017 INFO [decode.py:583] The transcripts are stored in pruned_transducer_stateless7_streaming_multi/exp/modified_beam_search/recogs-test-other-epoch-20-avg-4-streaming-chunk-size-32-modified_beam_search-beam-size-4-use-averaged-model.txt
2023-04-06 11:57:15,173 INFO [utils.py:558] [test-other-beam_size_4] %WER 5.96% [3118 / 52343, 346 ins, 280 del, 2492 sub ]
2023-04-06 11:57:15,529 INFO [decode.py:594] Wrote detailed error stats to pruned_transducer_stateless7_streaming_multi/exp/modified_beam_search/errs-test-other-epoch-20-avg-4-streaming-chunk-size-32-modified_beam_search-beam-size-4-use-averaged-model.txt
2023-04-06 11:57:15,529 INFO [decode.py:608] 
For test-other, WER of different settings are:
beam_size_4	5.96	best for test-other

2023-04-06 11:57:15,530 INFO [decode.py:814] Done!