decoding_results/greedy_search/log-decode-epoch-42-avg-6-context-1-max-sym-per-frame-1-use-averaged-model-2023-03-21-16-48-27

2023-03-21 16:48:27,403 INFO [decode.py:482] Decoding started
2023-03-21 16:48:27,403 INFO [decode.py:488] Device: cuda:0
2023-03-21 16:48:27,645 INFO [lexicon.py:168] Loading pre-compiled data/lang_char/Linv.pt
2023-03-21 16:48:27,670 INFO [decode.py:494] {'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': 'aishell_zipformer', 'icefall-git-sha1': '8337628-dirty', 'icefall-git-date': 'Mon Mar 20 16:20:05 2023', 'icefall-path': '/ceph-data4/yangxiaoyu/softwares/icefall_development/icefall_aishell_zipformer', '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-7-1218101249-5d97868c7c-v8ngc', 'IP address': '10.177.77.18'}, 'epoch': 42, 'iter': 0, 'avg': 6, 'use_averaged_model': True, 'exp_dir': PosixPath('pruned_transducer_stateless7/exp'), 'lang_dir': PosixPath('data/lang_char'), 'decoding_method': 'greedy_search', 'beam_size': 4, 'beam': 4, 'max_contexts': 4, 'max_states': 8, 'context_size': 1, 'max_sym_per_frame': 1, '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, '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_stateless7/exp/greedy_search'), 'suffix': 'epoch-42-avg-6-context-1-max-sym-per-frame-1-use-averaged-model', 'blank_id': 0, 'vocab_size': 4336}
2023-03-21 16:48:27,671 INFO [decode.py:496] About to create model
2023-03-21 16:48:28,361 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.
2023-03-21 16:48:28,460 INFO [decode.py:568] Calculating the averaged model over epoch range from 36 (excluded) to 42
2023-03-21 16:48:33,538 INFO [decode.py:591] Number of model parameters: 77741923
2023-03-21 16:48:33,539 INFO [aishell.py:51] About to get test cuts from data/fbank/aishell_cuts_test.jsonl.gz
2023-03-21 16:48:33,541 INFO [aishell.py:45] About to get valid cuts from data/fbank/aishell_cuts_dev.jsonl.gz
2023-03-21 16:48:37,730 INFO [decode.py:398] batch 0/?, cuts processed until now is 95
2023-03-21 16:48:47,465 INFO [zipformer.py:1455] attn_weights_entropy = tensor([4.1485, 4.4648, 4.1247, 4.3739, 3.9270, 4.4197, 4.6020, 4.6753],
       device='cuda:0'), covar=tensor([0.0192, 0.0104, 0.0187, 0.0150, 0.0288, 0.0183, 0.0200, 0.0139],
       device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0124, 0.0118, 0.0122, 0.0112, 0.0100, 0.0097, 0.0099],
       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')
2023-03-21 16:48:56,058 INFO [zipformer.py:1455] attn_weights_entropy = tensor([2.5963, 2.8684, 2.6166, 2.7795, 2.8613, 2.4702, 2.7993, 2.6541],
       device='cuda:0'), covar=tensor([0.0984, 0.0509, 0.1008, 0.0947, 0.0580, 0.0780, 0.0669, 0.1205],
       device='cuda:0'), in_proj_covar=tensor([0.0057, 0.0058, 0.0066, 0.0058, 0.0056, 0.0061, 0.0055, 0.0052],
       device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002],
       device='cuda:0')
2023-03-21 16:48:56,942 INFO [decode.py:398] batch 50/?, cuts processed until now is 5273
2023-03-21 16:49:02,930 INFO [decode.py:412] The transcripts are stored in pruned_transducer_stateless7/exp/greedy_search/recogs-test-epoch-42-avg-6-context-1-max-sym-per-frame-1-use-averaged-model.txt
2023-03-21 16:49:03,302 INFO [utils.py:558] [test-greedy_search] %WER 5.02% [5255 / 104765, 176 ins, 685 del, 4394 sub ]
2023-03-21 16:49:03,769 INFO [decode.py:427] Wrote detailed error stats to pruned_transducer_stateless7/exp/greedy_search/errs-test-epoch-42-avg-6-context-1-max-sym-per-frame-1-use-averaged-model.txt
2023-03-21 16:49:03,770 INFO [decode.py:441] 
For test, CER of different settings are:
greedy_search	5.02	best for test

2023-03-21 16:49:05,241 INFO [decode.py:398] batch 0/?, cuts processed until now is 107
2023-03-21 16:49:24,083 INFO [decode.py:398] batch 50/?, cuts processed until now is 6037
2023-03-21 16:49:42,427 INFO [decode.py:398] batch 100/?, cuts processed until now is 12026
2023-03-21 16:49:49,481 INFO [decode.py:412] The transcripts are stored in pruned_transducer_stateless7/exp/greedy_search/recogs-dev-epoch-42-avg-6-context-1-max-sym-per-frame-1-use-averaged-model.txt
2023-03-21 16:49:50,128 INFO [utils.py:558] [dev-greedy_search] %WER 4.61% [9472 / 205341, 368 ins, 1154 del, 7950 sub ]
2023-03-21 16:49:51,047 INFO [decode.py:427] Wrote detailed error stats to pruned_transducer_stateless7/exp/greedy_search/errs-dev-epoch-42-avg-6-context-1-max-sym-per-frame-1-use-averaged-model.txt
2023-03-21 16:49:51,048 INFO [decode.py:441] 
For dev, CER of different settings are:
greedy_search	4.61	best for dev

2023-03-21 16:49:51,064 INFO [decode.py:620] Done!