2024-06-24 10:30:31,308 INFO [decode.py:832] Decoding started
2024-06-24 10:30:31,309 INFO [decode.py:838] Device: cuda:0
2024-06-24 10:30:31,318 INFO [decode.py:848] {'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': '8f976a1e1407e330e2a233d68f81b1eb5269fdaa', 'k2-git-date': 'Thu Jun 6 02:13:08 2024', 'lhotse-version': '1.24.0.dev+git.4d57d53d.dirty', 'torch-version': '2.3.1+cu121', 'torch-cuda-available': True, 'torch-cuda-version': '12.1', 'python-version': '3.9', 'icefall-git-branch': 'feature/ksponspeech_zipformer', 'icefall-git-sha1': '7dda45c9-dirty', 'icefall-git-date': 'Tue Jun 18 16:40:30 2024', 'icefall-path': '/home/ubuntu/icefall', 'k2-path': '/home/ubuntu/miniforge3/envs/lhotse/lib/python3.9/site-packages/k2/__init__.py', 'lhotse-path': '/home/ubuntu/lhotse/lhotse/__init__.py', 'hostname': 'gpu-1', 'IP address': '127.0.1.1'}, 'epoch': 30, 'iter': 0, 'avg': 9, 'use_averaged_model': True, 'exp_dir': PosixPath('zipformer/exp'), 'bpe_model': 'data/lang_bpe_5000/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, 'use_shallow_fusion': False, 'lm_type': 'rnn', 'lm_scale': 0.3, 'tokens_ngram': 2, 'backoff_id': 500, 'context_score': 2, 'context_file': '', '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': False, '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, 'input_strategy': 'PrecomputedFeatures', 'lm_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('zipformer/exp/fast_beam_search'), 'has_contexts': False, 'suffix': 'epoch-30-avg-9-beam-20.0-max-contexts-8-max-states-64-use-averaged-model', 'blank_id': 0, 'unk_id': 2, 'vocab_size': 5000}
2024-06-24 10:30:31,318 INFO [decode.py:850] About to create model
2024-06-24 10:30:32,012 INFO [decode.py:917] Calculating the averaged model over epoch range from 21 (excluded) to 30
2024-06-24 10:30:37,819 INFO [decode.py:1011] Number of model parameters: 74778511
2024-06-24 10:30:37,819 INFO [asr_datamodule.py:405] About to get eval_clean cuts
2024-06-24 10:30:37,821 INFO [asr_datamodule.py:412] About to get eval_other cuts
2024-06-24 10:30:41,828 INFO [decode.py:705] batch 0/?, cuts processed until now is 21
2024-06-24 10:30:50,844 INFO [decode.py:705] batch 20/?, cuts processed until now is 1368
2024-06-24 10:30:56,165 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([3.6460, 3.3435, 2.4906, 3.5147], device='cuda:0')
2024-06-24 10:31:01,280 INFO [decode.py:705] batch 40/?, cuts processed until now is 2063
2024-06-24 10:31:10,045 INFO [decode.py:721] The transcripts are stored in zipformer/exp/fast_beam_search/recogs-eval_clean-beam_20.0_max_contexts_8_max_states_64-epoch-30-avg-9-beam-20.0-max-contexts-8-max-states-64-use-averaged-model.txt
2024-06-24 10:31:10,135 INFO [utils.py:656] [eval_clean-beam_20.0_max_contexts_8_max_states_64] %WER 10.59% [5134 / 48463, 951 ins, 1838 del, 2345 sub ]
2024-06-24 10:31:10,326 INFO [decode.py:734] Wrote detailed error stats to zipformer/exp/fast_beam_search/errs-eval_clean-beam_20.0_max_contexts_8_max_states_64-epoch-30-avg-9-beam-20.0-max-contexts-8-max-states-64-use-averaged-model.txt
2024-06-24 10:31:10,329 INFO [decode.py:750] 
For eval_clean, CER of different settings are:
beam_20.0_max_contexts_8_max_states_64	10.59	best for eval_clean

2024-06-24 10:31:12,334 INFO [decode.py:705] batch 0/?, cuts processed until now is 17
2024-06-24 10:31:22,306 INFO [decode.py:705] batch 20/?, cuts processed until now is 989
2024-06-24 10:31:28,298 INFO [zipformer.py:1858] name=None, attn_weights_entropy = tensor([4.7227, 4.1740, 3.1463, 4.4510], device='cuda:0')
2024-06-24 10:31:33,710 INFO [decode.py:705] batch 40/?, cuts processed until now is 1761
2024-06-24 10:31:44,195 INFO [decode.py:705] batch 60/?, cuts processed until now is 2371
2024-06-24 10:31:54,865 INFO [decode.py:705] batch 80/?, cuts processed until now is 2803
2024-06-24 10:31:57,079 INFO [decode.py:721] The transcripts are stored in zipformer/exp/fast_beam_search/recogs-eval_other-beam_20.0_max_contexts_8_max_states_64-epoch-30-avg-9-beam-20.0-max-contexts-8-max-states-64-use-averaged-model.txt
2024-06-24 10:31:57,340 INFO [utils.py:656] [eval_other-beam_20.0_max_contexts_8_max_states_64] %WER 11.54% [8202 / 71101, 1439 ins, 2941 del, 3822 sub ]
2024-06-24 10:31:57,629 INFO [decode.py:734] Wrote detailed error stats to zipformer/exp/fast_beam_search/errs-eval_other-beam_20.0_max_contexts_8_max_states_64-epoch-30-avg-9-beam-20.0-max-contexts-8-max-states-64-use-averaged-model.txt
2024-06-24 10:31:57,632 INFO [decode.py:750] 
For eval_other, CER of different settings are:
beam_20.0_max_contexts_8_max_states_64	11.54	best for eval_other

2024-06-24 10:31:57,635 INFO [decode.py:1046] Done!