File size: 5,108 Bytes
f8bdf58 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 |
2022-06-19 12:31:39,251 INFO [decode.py:497] Decoding started
2022-06-19 12:31:39,252 INFO [decode.py:503] Device: cuda:0
2022-06-19 12:31:39,489 INFO [lexicon.py:176] Loading pre-compiled data/lang_char/Linv.pt
2022-06-19 12:31:39,517 INFO [decode.py:509] {'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': 1000, 'feature_dim': 80, 'subsampling_factor': 4, 'model_warm_step': 3000, 'env_info': {'k2-version': '1.15.1', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': 'f8d2dba06c000ffee36aab5b66f24e7c9809f116', 'k2-git-date': 'Thu Apr 21 12:20:34 2022', 'lhotse-version': '1.3.0.dev+missing.version.file', 'torch-version': '1.10.0+cu102', 'torch-cuda-available': True, 'torch-cuda-version': '10.2', 'python-version': '3.8', 'icefall-git-branch': 'pruned-rnnt-aishell', 'icefall-git-sha1': 'd0a5f1d-dirty', 'icefall-git-date': 'Mon Jun 13 20:40:46 2022', 'icefall-path': '/k2-dev/fangjun/open-source/icefall-aishell', 'k2-path': '/ceph-fj/fangjun/open-source-2/k2-multi-22/k2/python/k2/__init__.py', 'lhotse-path': '/ceph-fj/fangjun/open-source-2/lhotse-jsonl/lhotse/__init__.py', 'hostname': 'de-74279-k2-train-7-0616225511-78bf4545d8-tv52r', 'IP address': '10.177.77.9'}, 'epoch': 29, 'iter': 0, 'avg': 5, 'use_averaged_model': True, 'exp_dir': PosixPath('pruned_transducer_stateless3/exp-context-size-1'), 'lang_dir': PosixPath('data/lang_char'), 'decoding_method': 'modified_beam_search', 'beam_size': 4, 'beam': 4, 'max_contexts': 4, 'max_states': 8, 'context_size': 1, 'max_sym_per_frame': 1, 'num_encoder_layers': 12, 'dim_feedforward': 2048, 'nhead': 8, 'encoder_dim': 512, 'decoder_dim': 512, 'joiner_dim': 512, 'max_duration': 600, 'bucketing_sampler': True, 'num_buckets': 30, 'shuffle': True, 'return_cuts': True, 'num_workers': 2, '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_stateless3/exp-context-size-1/modified_beam_search'), 'suffix': 'epoch-29-avg-5-modified_beam_search-beam-size-4-use-averaged-model', 'blank_id': 0, 'vocab_size': 4336}
2022-06-19 12:31:39,518 INFO [decode.py:511] About to create model
2022-06-19 12:31:40,323 INFO [decode.py:583] Calculating the averaged model over epoch range from 24 (excluded) to 29
2022-06-19 12:31:48,086 INFO [decode.py:606] Number of model parameters: 96983734
2022-06-19 12:31:48,087 INFO [aishell.py:51] About to get test cuts from data/fbank/aishell_cuts_test.jsonl.gz
2022-06-19 12:31:48,092 INFO [aishell.py:45] About to get valid cuts from data/fbank/aishell_cuts_dev.jsonl.gz
2022-06-19 12:31:54,287 INFO [decode.py:403] batch 0/?, cuts processed until now is 99
2022-06-19 12:33:25,765 INFO [decode.py:403] batch 20/?, cuts processed until now is 2264
2022-06-19 12:34:56,485 INFO [decode.py:403] batch 40/?, cuts processed until now is 4638
2022-06-19 12:36:23,030 INFO [decode.py:403] batch 60/?, cuts processed until now is 7129
2022-06-19 12:36:26,248 INFO [decode.py:420] The transcripts are stored in pruned_transducer_stateless3/exp-context-size-1/modified_beam_search/recogs-test-beam_size_4-epoch-29-avg-5-modified_beam_search-beam-size-4-use-averaged-model.txt
2022-06-19 12:36:26,625 INFO [utils.py:408] [test-beam_size_4] %WER 5.05% [5286 / 104765, 249 ins, 416 del, 4621 sub ]
2022-06-19 12:36:27,146 INFO [decode.py:437] Wrote detailed error stats to pruned_transducer_stateless3/exp-context-size-1/modified_beam_search/errs-test-beam_size_4-epoch-29-avg-5-modified_beam_search-beam-size-4-use-averaged-model.txt
2022-06-19 12:36:27,150 INFO [decode.py:454]
For test, CER of different settings are:
beam_size_4 5.05 best for test
2022-06-19 12:36:33,095 INFO [decode.py:403] batch 0/?, cuts processed until now is 111
2022-06-19 12:38:05,959 INFO [decode.py:403] batch 20/?, cuts processed until now is 2529
2022-06-19 12:39:38,100 INFO [decode.py:403] batch 40/?, cuts processed until now is 5042
2022-06-19 12:41:11,606 INFO [decode.py:403] batch 60/?, cuts processed until now is 7804
2022-06-19 12:42:44,331 INFO [decode.py:403] batch 80/?, cuts processed until now is 10338
2022-06-19 12:44:17,078 INFO [decode.py:403] batch 100/?, cuts processed until now is 13159
2022-06-19 12:44:54,956 INFO [decode.py:420] The transcripts are stored in pruned_transducer_stateless3/exp-context-size-1/modified_beam_search/recogs-dev-beam_size_4-epoch-29-avg-5-modified_beam_search-beam-size-4-use-averaged-model.txt
2022-06-19 12:44:55,370 INFO [utils.py:408] [dev-beam_size_4] %WER 4.79% [9846 / 205341, 444 ins, 703 del, 8699 sub ]
2022-06-19 12:44:56,305 INFO [decode.py:437] Wrote detailed error stats to pruned_transducer_stateless3/exp-context-size-1/modified_beam_search/errs-dev-beam_size_4-epoch-29-avg-5-modified_beam_search-beam-size-4-use-averaged-model.txt
2022-06-19 12:44:56,306 INFO [decode.py:454]
For dev, CER of different settings are:
beam_size_4 4.79 best for dev
2022-06-19 12:44:56,323 INFO [decode.py:633] Done!
|