decoding_results/fast_beam_search/log-decode-epoch-30-avg-9-streaming-chunk-size-32-beam-20.0-max-contexts-8-max-states-64-use-averaged-model-2022-12-26-15-53-41

2022-12-26 15:53:41,676 INFO [decode.py:655] Decoding started
2022-12-26 15:53:41,676 INFO [decode.py:661] Device: cuda:0
2022-12-26 15:53:41,682 INFO [decode.py:671] {'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': '1.23.2', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': 'efd83642a940dc7db08688cc0791985bed1fafcd', 'k2-git-date': 'Sun Nov 27 19:12:00 2022', 'lhotse-version': '1.12.0.dev+git.891bad1.clean', 'torch-version': '1.10.0+cu102', 'torch-cuda-available': True, 'torch-cuda-version': '10.2', 'python-version': '3.8', 'icefall-git-branch': 'streaming_zipformer', 'icefall-git-sha1': '3c5ed61-clean', 'icefall-git-date': 'Mon Dec 26 13:01:22 2022', 'icefall-path': '/star-zw/workspace/zipformer/icefall_streaming2', 'k2-path': '/star-zw/workspace/share/k2-last/k2/python/k2/__init__.py', 'lhotse-path': '/star-zw/env/k2_icefall/lib/python3.8/site-packages/lhotse-1.12.0.dev0+git.891bad1.clean-py3.8.egg/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-2-1216192652-5bcf7587b4-n6q9m', 'IP address': '10.177.74.211'}, 'epoch': 30, 'iter': 0, 'avg': 9, 'use_averaged_model': True, 'exp_dir': PosixPath('pruned_transducer_stateless7_streaming/exp-full-dynamic-chunk'), 'bpe_model': 'data/lang_bpe_500/bpe.model', 'lang_dir': PosixPath('data/lang_bpe_500'), 'decoding_method': 'fast_beam_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': '2,4,3,2,4', '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, 'full_libri': True, 'manifest_dir': PosixPath('data/fbank'), 'max_duration': 600, '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('pruned_transducer_stateless7_streaming/exp-full-dynamic-chunk/fast_beam_search'), 'suffix': 'epoch-30-avg-9-streaming-chunk-size-32-beam-20.0-max-contexts-8-max-states-64-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 500}
2022-12-26 15:53:41,682 INFO [decode.py:673] About to create model
2022-12-26 15:53:42,327 INFO [zipformer.py:378] 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.
2022-12-26 15:53:42,341 INFO [decode.py:744] Calculating the averaged model over epoch range from 21 (excluded) to 30
2022-12-26 15:53:56,411 INFO [decode.py:778] Number of model parameters: 70369391
2022-12-26 15:53:56,411 INFO [asr_datamodule.py:443] About to get test-clean cuts
2022-12-26 15:53:56,421 INFO [asr_datamodule.py:450] About to get test-other cuts
2022-12-26 15:53:59,270 INFO [decode.py:560] batch 0/?, cuts processed until now is 43
2022-12-26 15:54:31,199 INFO [decode.py:560] batch 20/?, cuts processed until now is 1430
2022-12-26 15:54:55,125 INFO [zipformer.py:2453] attn_weights_entropy = tensor([2.6072, 3.4398, 3.2798, 1.3989, 3.5262, 2.7791, 0.6962, 2.0624],
       device='cuda:0'), covar=tensor([0.1968, 0.0881, 0.1401, 0.4125, 0.0731, 0.0908, 0.4646, 0.1797],
       device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0144, 0.0158, 0.0122, 0.0147, 0.0112, 0.0142, 0.0113],
       device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002],
       device='cuda:0')
2022-12-26 15:55:04,830 INFO [decode.py:560] batch 40/?, cuts processed until now is 2561
2022-12-26 15:55:05,057 INFO [zipformer.py:2453] attn_weights_entropy = tensor([1.9880, 1.2823, 1.9022, 1.6360, 2.0849, 2.0807, 1.7661, 1.8778],
       device='cuda:0'), covar=tensor([0.2520, 0.3568, 0.2734, 0.3188, 0.2412, 0.1061, 0.3954, 0.1427],
       device='cuda:0'), in_proj_covar=tensor([0.0267, 0.0297, 0.0281, 0.0320, 0.0312, 0.0254, 0.0348, 0.0243],
       device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002],
       device='cuda:0')
2022-12-26 15:55:05,901 INFO [decode.py:576] The transcripts are stored in pruned_transducer_stateless7_streaming/exp-full-dynamic-chunk/fast_beam_search/recogs-test-clean-beam_20.0_max_contexts_8_max_states_64-epoch-30-avg-9-streaming-chunk-size-32-beam-20.0-max-contexts-8-max-states-64-use-averaged-model.txt
2022-12-26 15:55:05,977 INFO [utils.py:536] [test-clean-beam_20.0_max_contexts_8_max_states_64] %WER 3.20% [1685 / 52576, 196 ins, 126 del, 1363 sub ]
2022-12-26 15:55:06,154 INFO [decode.py:589] Wrote detailed error stats to pruned_transducer_stateless7_streaming/exp-full-dynamic-chunk/fast_beam_search/errs-test-clean-beam_20.0_max_contexts_8_max_states_64-epoch-30-avg-9-streaming-chunk-size-32-beam-20.0-max-contexts-8-max-states-64-use-averaged-model.txt
2022-12-26 15:55:06,157 INFO [decode.py:605] 
For test-clean, WER of different settings are:
beam_20.0_max_contexts_8_max_states_64	3.2	best for test-clean

2022-12-26 15:55:08,413 INFO [decode.py:560] batch 0/?, cuts processed until now is 52
2022-12-26 15:55:19,276 INFO [zipformer.py:2453] attn_weights_entropy = tensor([1.5789, 1.4018, 1.3338, 1.7230, 1.7707, 2.9352, 1.5062, 1.5780],
       device='cuda:0'), covar=tensor([0.0825, 0.1946, 0.0994, 0.0844, 0.1375, 0.0310, 0.1411, 0.1567],
       device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0081, 0.0070, 0.0073, 0.0089, 0.0074, 0.0083, 0.0076],
       device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004],
       device='cuda:0')
2022-12-26 15:55:38,950 INFO [decode.py:560] batch 20/?, cuts processed until now is 1647
2022-12-26 15:56:11,524 INFO [decode.py:560] batch 40/?, cuts processed until now is 2870
2022-12-26 15:56:12,579 INFO [decode.py:576] The transcripts are stored in pruned_transducer_stateless7_streaming/exp-full-dynamic-chunk/fast_beam_search/recogs-test-other-beam_20.0_max_contexts_8_max_states_64-epoch-30-avg-9-streaming-chunk-size-32-beam-20.0-max-contexts-8-max-states-64-use-averaged-model.txt
2022-12-26 15:56:12,658 INFO [utils.py:536] [test-other-beam_20.0_max_contexts_8_max_states_64] %WER 8.04% [4210 / 52343, 428 ins, 400 del, 3382 sub ]
2022-12-26 15:56:12,924 INFO [decode.py:589] Wrote detailed error stats to pruned_transducer_stateless7_streaming/exp-full-dynamic-chunk/fast_beam_search/errs-test-other-beam_20.0_max_contexts_8_max_states_64-epoch-30-avg-9-streaming-chunk-size-32-beam-20.0-max-contexts-8-max-states-64-use-averaged-model.txt
2022-12-26 15:56:12,927 INFO [decode.py:605] 
For test-other, WER of different settings are:
beam_20.0_max_contexts_8_max_states_64	8.04	best for test-other

2022-12-26 15:56:12,928 INFO [decode.py:809] Done!