Hugging Face's logo

bookbot
/
zipformer-streaming-robust-es-v0

Automatic Speech Recognition
icefall
phoneme-recognition
Model card Files
xet
 Community
Davidsamuel101 commited on
Commit
5db1419
·
1 Parent(s): 5527fc8

Commit model

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. data/lang_phone/tokens.txt +37 -0
  2. epoch-76.pt +3 -0
  3. epoch-77.pt +3 -0
  4. epoch-78.pt +3 -0
  5. epoch-79.pt +3 -0
  6. epoch-80.pt +3 -0
  7. streaming/fast_beam_search/errs-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt +0 -0
  8. streaming/fast_beam_search/errs-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt +0 -0
  9. streaming/fast_beam_search/errs-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt +0 -0
  10. streaming/fast_beam_search/errs-test-slr72-epoch-80-avg-5-chunk-32-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt +0 -0
  11. streaming/fast_beam_search/log-decode-epoch-80-avg-5-chunk-16-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model-2025-11-17-17-41-10 +288 -0
  12. streaming/fast_beam_search/log-decode-epoch-80-avg-5-chunk-32-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model-2025-11-17-16-49-23 +114 -0
  13. streaming/fast_beam_search/log-decode-epoch-80-avg-5-chunk-32-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model-2025-11-17-16-58-25 +288 -0
  14. streaming/fast_beam_search/recogs-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt +0 -0
  15. streaming/fast_beam_search/recogs-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt +0 -0
  16. streaming/fast_beam_search/recogs-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt +0 -0
  17. streaming/fast_beam_search/recogs-test-slr72-epoch-80-avg-5-chunk-32-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt +0 -0
  18. streaming/fast_beam_search/wer-summary-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt +2 -0
  19. streaming/fast_beam_search/wer-summary-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt +2 -0
  20. streaming/fast_beam_search/wer-summary-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt +2 -0
  21. streaming/fast_beam_search/wer-summary-test-slr72-epoch-80-avg-5-chunk-32-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt +2 -0
  22. streaming/greedy_search/errs-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt +0 -0
  23. streaming/greedy_search/errs-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt +0 -0
  24. streaming/greedy_search/errs-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt +0 -0
  25. streaming/greedy_search/errs-test-slr72-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt +0 -0
  26. streaming/greedy_search/log-decode-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model-2025-11-17-17-37-14 +288 -0
  27. streaming/greedy_search/log-decode-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model-2025-11-17-16-48-52 +114 -0
  28. streaming/greedy_search/log-decode-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model-2025-11-17-16-55-04 +288 -0
  29. streaming/greedy_search/recogs-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt +0 -0
  30. streaming/greedy_search/recogs-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt +0 -0
  31. streaming/greedy_search/recogs-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt +0 -0
  32. streaming/greedy_search/recogs-test-slr72-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt +0 -0
  33. streaming/greedy_search/wer-summary-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt +2 -0
  34. streaming/greedy_search/wer-summary-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt +2 -0
  35. streaming/greedy_search/wer-summary-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt +2 -0
  36. streaming/greedy_search/wer-summary-test-slr72-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt +2 -0
  37. streaming/modified_beam_search/errs-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt +0 -0
  38. streaming/modified_beam_search/errs-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt +0 -0
  39. streaming/modified_beam_search/errs-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt +0 -0
  40. streaming/modified_beam_search/errs-test-slr72-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt +0 -0
  41. streaming/modified_beam_search/log-decode-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model-2025-11-17-17-47-46 +288 -0
  42. streaming/modified_beam_search/log-decode-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model-2025-11-17-16-49-52 +114 -0
  43. streaming/modified_beam_search/log-decode-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model-2025-11-17-17-03-31 +288 -0
  44. streaming/modified_beam_search/recogs-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt +0 -0
  45. streaming/modified_beam_search/recogs-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt +0 -0
  46. streaming/modified_beam_search/recogs-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt +0 -0
  47. streaming/modified_beam_search/recogs-test-slr72-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt +0 -0
  48. streaming/modified_beam_search/wer-summary-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt +2 -0
  49. streaming/modified_beam_search/wer-summary-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt +2 -0
  50. streaming/modified_beam_search/wer-summary-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt +2 -0
data/lang_phone/tokens.txt ADDED
@@ -0,0 +1,37 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ <eps> 0
2
+ <UNK> 1
3
+ a 2
4
+ ai 3
5
+ au 4
6
+ b 5
7
+ d 6
8
+ e 7
9
+ ei 8
10
+ eu 9
11
+ f 10
12
+ g 11
13
+ i 12
14
+ k 13
15
+ l 14
16
+ m 15
17
+ n 16
18
+ o 17
19
+ oi 18
20
+ ou 19
21
+ p 20
22
+ r 21
23
+ s 22
24
+ t 23
25
+ t͡ʃ 24
26
+ u 25
27
+ wa 26
28
+ we 27
29
+ wi 28
30
+ wo 29
31
+ x 30
32
+ ɲ 31
33
+ ɾ 32
34
+ ʎ 33
35
+ ʝ 34
36
+ θ 35
37
+ #0 36
epoch-76.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:feeda08a77e4d7ba33e41a85797a993876d0f2eaedec23b45d596c747c28bfdb
3
+ size 365539229
epoch-77.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e5d7f055fcf5ba6eca2b0231c218a110ce307dec0e2a2413d032a847334b2d1d
3
+ size 365539293
epoch-78.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:89f64e0d1e4be49c519c8acba58f8cfe3b19815bed139f45ce8e3ccff8fa78bb
3
+ size 365539357
epoch-79.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c3c52f31665578d001fcf76cdb0b61714232de0639052ea047e29b38a59d05c0
3
+ size 365539421
epoch-80.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dfc84e3c7479eac01aa56773ce217895270fc7acf5ca4d30c77f151f49dcbc3e
3
+ size 365539421
streaming/fast_beam_search/errs-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/fast_beam_search/errs-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/fast_beam_search/errs-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/fast_beam_search/errs-test-slr72-epoch-80-avg-5-chunk-32-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/fast_beam_search/log-decode-epoch-80-avg-5-chunk-16-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model-2025-11-17-17-41-10 ADDED
@@ -0,0 +1,288 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2025-11-17 17:41:10,972 INFO [streaming_decode.py:677] Decoding started
2
+ 2025-11-17 17:41:10,972 INFO [streaming_decode.py:683] Device: cuda:0
3
+ 2025-11-17 17:41:10,972 INFO [lexicon.py:168] Loading pre-compiled data/lang_phone/Linv.pt
4
+ 2025-11-17 17:41:10,973 INFO [streaming_decode.py:691] {
5
+ "attention_decoder_attention_dim": 512,
6
+ "attention_decoder_dim": 512,
7
+ "attention_decoder_feedforward_dim": 2048,
8
+ "attention_decoder_num_heads": 8,
9
+ "attention_decoder_num_layers": 6,
10
+ "avg": 5,
11
+ "batch_idx_train": 0,
12
+ "beam": 4,
13
+ "best_train_epoch": -1,
14
+ "best_train_loss": Infinity,
15
+ "best_valid_epoch": -1,
16
+ "best_valid_loss": Infinity,
17
+ "blank_id": 0,
18
+ "bucketing_sampler": true,
19
+ "causal": true,
20
+ "chunk_size": "16",
21
+ "cnn_module_kernel": "31,31,15,15,15,31",
22
+ "concatenate_cuts": false,
23
+ "context_size": 2,
24
+ "decoder_dim": 512,
25
+ "decoding_method": "fast_beam_search",
26
+ "downsampling_factor": "1,2,4,8,4,2",
27
+ "drop_last": true,
28
+ "duration_factor": 1.0,
29
+ "enable_musan": true,
30
+ "enable_spec_aug": true,
31
+ "encoder_dim": "192,256,256,256,256,256",
32
+ "encoder_unmasked_dim": "192,192,192,192,192,192",
33
+ "env_info": {
34
+ "IP address": "127.0.1.1",
35
+ "hostname": "Bookbot-GPU1",
36
+ "icefall-git-branch": "master",
37
+ "icefall-git-date": "Tue Jan 7 16:20:44 2025",
38
+ "icefall-git-sha1": "77cd018f-dirty",
39
+ "icefall-path": "/mnt/Projects/Projects/ASR/icefall",
40
+ "k2-build-type": "Release",
41
+ "k2-git-date": "Thu Jul 25 03:46:03 2024",
42
+ "k2-git-sha1": "5735fa707f6091856d13ccd230aced6e9e64f815",
43
+ "k2-path": "/home/bookbot/miniconda3/envs/icefall/lib/python3.12/site-packages/k2/__init__.py",
44
+ "k2-version": "1.24.4",
45
+ "k2-with-cuda": true,
46
+ "lhotse-path": "/home/bookbot/miniconda3/envs/icefall/lib/python3.12/site-packages/lhotse/__init__.py",
47
+ "lhotse-version": "1.31.1.dev+git.4c43958.clean",
48
+ "python-version": "3.12",
49
+ "torch-cuda-available": true,
50
+ "torch-cuda-version": "12.4",
51
+ "torch-version": "2.4.0+cu124"
52
+ },
53
+ "epoch": 80,
54
+ "exp_dir": "tmp/exp-causal-80-epoch",
55
+ "feature_dim": 80,
56
+ "feedforward_dim": "512,768,768,768,768,768",
57
+ "gap": 1.0,
58
+ "ignore_id": -1,
59
+ "input_strategy": "PrecomputedFeatures",
60
+ "iter": 0,
61
+ "joiner_dim": 512,
62
+ "label_smoothing": 0.1,
63
+ "lang_dir": "data/lang_phone",
64
+ "left_context_frames": "128",
65
+ "log_interval": 50,
66
+ "manifest_dir": "data/fbank",
67
+ "max_contexts": 4,
68
+ "max_duration": 200.0,
69
+ "max_states": 32,
70
+ "num_active_paths": 4,
71
+ "num_buckets": 30,
72
+ "num_decode_streams": 1000,
73
+ "num_encoder_layers": "2,2,2,2,2,2",
74
+ "num_heads": "4,4,4,8,4,4",
75
+ "num_workers": 2,
76
+ "on_the_fly_feats": false,
77
+ "pos_dim": 48,
78
+ "pos_head_dim": "4",
79
+ "query_head_dim": "32",
80
+ "res_dir": "tmp/exp-causal-80-epoch/streaming/fast_beam_search",
81
+ "reset_interval": 200,
82
+ "return_cuts": true,
83
+ "shuffle": true,
84
+ "spec_aug_time_warp_factor": 80,
85
+ "subsampling_factor": 4,
86
+ "suffix": "epoch-80-avg-5-chunk-16-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model",
87
+ "unk_id": 1,
88
+ "use_attention_decoder": false,
89
+ "use_averaged_model": true,
90
+ "use_cr_ctc": false,
91
+ "use_ctc": false,
92
+ "use_transducer": true,
93
+ "valid_interval": 3000,
94
+ "value_head_dim": "12",
95
+ "vocab_size": 36,
96
+ "warm_step": 2000
97
+ }
98
+ 2025-11-17 17:41:10,973 INFO [streaming_decode.py:693] About to create model
99
+ 2025-11-17 17:41:11,139 INFO [streaming_decode.py:748] Calculating the averaged model over epoch range from 75 (excluded) to 80
100
+ 2025-11-17 17:41:12,171 INFO [streaming_decode.py:769] Number of model parameters: 22795007
101
+ 2025-11-17 17:41:12,171 INFO [multidataset.py:83] About to get Spanish Common Voice test cuts
102
+ 2025-11-17 17:41:12,171 INFO [multidataset.py:85] Loading Spanish Common Voice in lazy mode
103
+ 2025-11-17 17:41:12,172 INFO [multidataset.py:94] About to get SLR72 dataset test cuts
104
+ 2025-11-17 17:41:12,172 INFO [multidataset.py:96] Loading SLR72 dataset in lazy mode
105
+ 2025-11-17 17:41:12,275 INFO [streaming_decode.py:582] Cuts processed until now is 0.
106
+ 2025-11-17 17:41:13,184 INFO [streaming_decode.py:582] Cuts processed until now is 100.
107
+ 2025-11-17 17:41:13,861 INFO [streaming_decode.py:582] Cuts processed until now is 200.
108
+ 2025-11-17 17:41:14,617 INFO [streaming_decode.py:582] Cuts processed until now is 300.
109
+ 2025-11-17 17:41:15,413 INFO [streaming_decode.py:582] Cuts processed until now is 400.
110
+ 2025-11-17 17:41:16,334 INFO [streaming_decode.py:582] Cuts processed until now is 500.
111
+ 2025-11-17 17:41:17,201 INFO [streaming_decode.py:582] Cuts processed until now is 600.
112
+ 2025-11-17 17:41:17,977 INFO [streaming_decode.py:582] Cuts processed until now is 700.
113
+ 2025-11-17 17:41:18,869 INFO [streaming_decode.py:582] Cuts processed until now is 800.
114
+ 2025-11-17 17:41:19,595 INFO [streaming_decode.py:582] Cuts processed until now is 900.
115
+ 2025-11-17 17:41:25,418 INFO [streaming_decode.py:582] Cuts processed until now is 1000.
116
+ 2025-11-17 17:41:30,171 INFO [streaming_decode.py:582] Cuts processed until now is 1100.
117
+ 2025-11-17 17:41:32,521 INFO [streaming_decode.py:582] Cuts processed until now is 1200.
118
+ 2025-11-17 17:41:34,493 INFO [streaming_decode.py:582] Cuts processed until now is 1300.
119
+ 2025-11-17 17:41:36,098 INFO [streaming_decode.py:582] Cuts processed until now is 1400.
120
+ 2025-11-17 17:41:37,597 INFO [streaming_decode.py:582] Cuts processed until now is 1500.
121
+ 2025-11-17 17:41:39,793 INFO [streaming_decode.py:582] Cuts processed until now is 1600.
122
+ 2025-11-17 17:41:41,095 INFO [streaming_decode.py:582] Cuts processed until now is 1700.
123
+ 2025-11-17 17:41:43,509 INFO [streaming_decode.py:582] Cuts processed until now is 1800.
124
+ 2025-11-17 17:41:45,608 INFO [streaming_decode.py:582] Cuts processed until now is 1900.
125
+ 2025-11-17 17:41:48,943 INFO [streaming_decode.py:582] Cuts processed until now is 2000.
126
+ 2025-11-17 17:41:51,984 INFO [streaming_decode.py:582] Cuts processed until now is 2100.
127
+ 2025-11-17 17:41:54,272 INFO [streaming_decode.py:582] Cuts processed until now is 2200.
128
+ 2025-11-17 17:41:57,078 INFO [streaming_decode.py:582] Cuts processed until now is 2300.
129
+ 2025-11-17 17:41:58,778 INFO [streaming_decode.py:582] Cuts processed until now is 2400.
130
+ 2025-11-17 17:42:01,104 INFO [streaming_decode.py:582] Cuts processed until now is 2500.
131
+ 2025-11-17 17:42:03,118 INFO [streaming_decode.py:582] Cuts processed until now is 2600.
132
+ 2025-11-17 17:42:05,408 INFO [streaming_decode.py:582] Cuts processed until now is 2700.
133
+ 2025-11-17 17:42:07,596 INFO [streaming_decode.py:582] Cuts processed until now is 2800.
134
+ 2025-11-17 17:42:09,700 INFO [streaming_decode.py:582] Cuts processed until now is 2900.
135
+ 2025-11-17 17:42:11,930 INFO [streaming_decode.py:582] Cuts processed until now is 3000.
136
+ 2025-11-17 17:42:14,796 INFO [streaming_decode.py:582] Cuts processed until now is 3100.
137
+ 2025-11-17 17:42:17,077 INFO [streaming_decode.py:582] Cuts processed until now is 3200.
138
+ 2025-11-17 17:42:19,285 INFO [streaming_decode.py:582] Cuts processed until now is 3300.
139
+ 2025-11-17 17:42:21,469 INFO [streaming_decode.py:582] Cuts processed until now is 3400.
140
+ 2025-11-17 17:42:23,650 INFO [streaming_decode.py:582] Cuts processed until now is 3500.
141
+ 2025-11-17 17:42:25,746 INFO [streaming_decode.py:582] Cuts processed until now is 3600.
142
+ 2025-11-17 17:42:27,944 INFO [streaming_decode.py:582] Cuts processed until now is 3700.
143
+ 2025-11-17 17:42:30,113 INFO [streaming_decode.py:582] Cuts processed until now is 3800.
144
+ 2025-11-17 17:42:32,906 INFO [streaming_decode.py:582] Cuts processed until now is 3900.
145
+ 2025-11-17 17:42:34,730 INFO [streaming_decode.py:582] Cuts processed until now is 4000.
146
+ 2025-11-17 17:42:36,521 INFO [streaming_decode.py:582] Cuts processed until now is 4100.
147
+ 2025-11-17 17:42:38,896 INFO [streaming_decode.py:582] Cuts processed until now is 4200.
148
+ 2025-11-17 17:42:41,014 INFO [streaming_decode.py:582] Cuts processed until now is 4300.
149
+ 2025-11-17 17:42:43,220 INFO [streaming_decode.py:582] Cuts processed until now is 4400.
150
+ 2025-11-17 17:42:45,588 INFO [streaming_decode.py:582] Cuts processed until now is 4500.
151
+ 2025-11-17 17:42:47,710 INFO [streaming_decode.py:582] Cuts processed until now is 4600.
152
+ 2025-11-17 17:42:49,781 INFO [streaming_decode.py:582] Cuts processed until now is 4700.
153
+ 2025-11-17 17:42:51,923 INFO [streaming_decode.py:582] Cuts processed until now is 4800.
154
+ 2025-11-17 17:42:54,122 INFO [streaming_decode.py:582] Cuts processed until now is 4900.
155
+ 2025-11-17 17:42:56,323 INFO [streaming_decode.py:582] Cuts processed until now is 5000.
156
+ 2025-11-17 17:42:58,564 INFO [streaming_decode.py:582] Cuts processed until now is 5100.
157
+ 2025-11-17 17:43:01,401 INFO [streaming_decode.py:582] Cuts processed until now is 5200.
158
+ 2025-11-17 17:43:03,452 INFO [streaming_decode.py:582] Cuts processed until now is 5300.
159
+ 2025-11-17 17:43:05,687 INFO [streaming_decode.py:582] Cuts processed until now is 5400.
160
+ 2025-11-17 17:43:07,767 INFO [streaming_decode.py:582] Cuts processed until now is 5500.
161
+ 2025-11-17 17:43:09,258 INFO [streaming_decode.py:582] Cuts processed until now is 5600.
162
+ 2025-11-17 17:43:12,458 INFO [streaming_decode.py:582] Cuts processed until now is 5700.
163
+ 2025-11-17 17:43:14,765 INFO [streaming_decode.py:582] Cuts processed until now is 5800.
164
+ 2025-11-17 17:43:16,970 INFO [streaming_decode.py:582] Cuts processed until now is 5900.
165
+ 2025-11-17 17:43:19,335 INFO [streaming_decode.py:582] Cuts processed until now is 6000.
166
+ 2025-11-17 17:43:22,194 INFO [streaming_decode.py:582] Cuts processed until now is 6100.
167
+ 2025-11-17 17:43:24,277 INFO [streaming_decode.py:582] Cuts processed until now is 6200.
168
+ 2025-11-17 17:43:26,609 INFO [streaming_decode.py:582] Cuts processed until now is 6300.
169
+ 2025-11-17 17:43:28,854 INFO [streaming_decode.py:582] Cuts processed until now is 6400.
170
+ 2025-11-17 17:43:30,964 INFO [streaming_decode.py:582] Cuts processed until now is 6500.
171
+ 2025-11-17 17:43:32,843 INFO [streaming_decode.py:582] Cuts processed until now is 6600.
172
+ 2025-11-17 17:43:35,051 INFO [streaming_decode.py:582] Cuts processed until now is 6700.
173
+ 2025-11-17 17:43:37,359 INFO [streaming_decode.py:582] Cuts processed until now is 6800.
174
+ 2025-11-17 17:43:39,942 INFO [streaming_decode.py:582] Cuts processed until now is 6900.
175
+ 2025-11-17 17:43:42,566 INFO [streaming_decode.py:582] Cuts processed until now is 7000.
176
+ 2025-11-17 17:43:44,824 INFO [streaming_decode.py:582] Cuts processed until now is 7100.
177
+ 2025-11-17 17:43:47,794 INFO [streaming_decode.py:582] Cuts processed until now is 7200.
178
+ 2025-11-17 17:43:50,220 INFO [streaming_decode.py:582] Cuts processed until now is 7300.
179
+ 2025-11-17 17:43:52,193 INFO [streaming_decode.py:582] Cuts processed until now is 7400.
180
+ 2025-11-17 17:43:54,398 INFO [streaming_decode.py:582] Cuts processed until now is 7500.
181
+ 2025-11-17 17:43:56,738 INFO [streaming_decode.py:582] Cuts processed until now is 7600.
182
+ 2025-11-17 17:43:59,007 INFO [streaming_decode.py:582] Cuts processed until now is 7700.
183
+ 2025-11-17 17:44:01,441 INFO [streaming_decode.py:582] Cuts processed until now is 7800.
184
+ 2025-11-17 17:44:03,408 INFO [streaming_decode.py:582] Cuts processed until now is 7900.
185
+ 2025-11-17 17:44:05,303 INFO [streaming_decode.py:582] Cuts processed until now is 8000.
186
+ 2025-11-17 17:44:07,254 INFO [streaming_decode.py:582] Cuts processed until now is 8100.
187
+ 2025-11-17 17:44:09,508 INFO [streaming_decode.py:582] Cuts processed until now is 8200.
188
+ 2025-11-17 17:44:11,989 INFO [streaming_decode.py:582] Cuts processed until now is 8300.
189
+ 2025-11-17 17:44:14,290 INFO [streaming_decode.py:582] Cuts processed until now is 8400.
190
+ 2025-11-17 17:44:17,051 INFO [streaming_decode.py:582] Cuts processed until now is 8500.
191
+ 2025-11-17 17:44:19,787 INFO [streaming_decode.py:582] Cuts processed until now is 8600.
192
+ 2025-11-17 17:44:22,028 INFO [streaming_decode.py:582] Cuts processed until now is 8700.
193
+ 2025-11-17 17:44:24,344 INFO [streaming_decode.py:582] Cuts processed until now is 8800.
194
+ 2025-11-17 17:44:26,501 INFO [streaming_decode.py:582] Cuts processed until now is 8900.
195
+ 2025-11-17 17:44:28,735 INFO [streaming_decode.py:582] Cuts processed until now is 9000.
196
+ 2025-11-17 17:44:31,019 INFO [streaming_decode.py:582] Cuts processed until now is 9100.
197
+ 2025-11-17 17:44:33,089 INFO [streaming_decode.py:582] Cuts processed until now is 9200.
198
+ 2025-11-17 17:44:35,254 INFO [streaming_decode.py:582] Cuts processed until now is 9300.
199
+ 2025-11-17 17:44:37,271 INFO [streaming_decode.py:582] Cuts processed until now is 9400.
200
+ 2025-11-17 17:44:39,313 INFO [streaming_decode.py:582] Cuts processed until now is 9500.
201
+ 2025-11-17 17:44:42,160 INFO [streaming_decode.py:582] Cuts processed until now is 9600.
202
+ 2025-11-17 17:44:44,591 INFO [streaming_decode.py:582] Cuts processed until now is 9700.
203
+ 2025-11-17 17:44:46,783 INFO [streaming_decode.py:582] Cuts processed until now is 9800.
204
+ 2025-11-17 17:44:49,219 INFO [streaming_decode.py:582] Cuts processed until now is 9900.
205
+ 2025-11-17 17:44:51,750 INFO [streaming_decode.py:582] Cuts processed until now is 10000.
206
+ 2025-11-17 17:44:54,104 INFO [streaming_decode.py:582] Cuts processed until now is 10100.
207
+ 2025-11-17 17:44:56,545 INFO [streaming_decode.py:582] Cuts processed until now is 10200.
208
+ 2025-11-17 17:44:58,726 INFO [streaming_decode.py:582] Cuts processed until now is 10300.
209
+ 2025-11-17 17:45:00,644 INFO [streaming_decode.py:582] Cuts processed until now is 10400.
210
+ 2025-11-17 17:45:03,771 INFO [streaming_decode.py:582] Cuts processed until now is 10500.
211
+ 2025-11-17 17:45:05,362 INFO [streaming_decode.py:582] Cuts processed until now is 10600.
212
+ 2025-11-17 17:45:07,448 INFO [streaming_decode.py:582] Cuts processed until now is 10700.
213
+ 2025-11-17 17:45:10,045 INFO [streaming_decode.py:582] Cuts processed until now is 10800.
214
+ 2025-11-17 17:45:12,391 INFO [streaming_decode.py:582] Cuts processed until now is 10900.
215
+ 2025-11-17 17:45:15,418 INFO [streaming_decode.py:582] Cuts processed until now is 11000.
216
+ 2025-11-17 17:45:17,728 INFO [streaming_decode.py:582] Cuts processed until now is 11100.
217
+ 2025-11-17 17:45:20,001 INFO [streaming_decode.py:582] Cuts processed until now is 11200.
218
+ 2025-11-17 17:45:22,555 INFO [streaming_decode.py:582] Cuts processed until now is 11300.
219
+ 2025-11-17 17:45:24,820 INFO [streaming_decode.py:582] Cuts processed until now is 11400.
220
+ 2025-11-17 17:45:27,726 INFO [streaming_decode.py:582] Cuts processed until now is 11500.
221
+ 2025-11-17 17:45:30,077 INFO [streaming_decode.py:582] Cuts processed until now is 11600.
222
+ 2025-11-17 17:45:32,390 INFO [streaming_decode.py:582] Cuts processed until now is 11700.
223
+ 2025-11-17 17:45:34,714 INFO [streaming_decode.py:582] Cuts processed until now is 11800.
224
+ 2025-11-17 17:45:36,034 INFO [streaming_decode.py:582] Cuts processed until now is 11900.
225
+ 2025-11-17 17:45:38,878 INFO [streaming_decode.py:582] Cuts processed until now is 12000.
226
+ 2025-11-17 17:45:41,004 INFO [streaming_decode.py:582] Cuts processed until now is 12100.
227
+ 2025-11-17 17:45:43,294 INFO [streaming_decode.py:582] Cuts processed until now is 12200.
228
+ 2025-11-17 17:45:45,661 INFO [streaming_decode.py:582] Cuts processed until now is 12300.
229
+ 2025-11-17 17:45:47,865 INFO [streaming_decode.py:582] Cuts processed until now is 12400.
230
+ 2025-11-17 17:45:50,107 INFO [streaming_decode.py:582] Cuts processed until now is 12500.
231
+ 2025-11-17 17:45:52,950 INFO [streaming_decode.py:582] Cuts processed until now is 12600.
232
+ 2025-11-17 17:45:54,558 INFO [streaming_decode.py:582] Cuts processed until now is 12700.
233
+ 2025-11-17 17:45:56,798 INFO [streaming_decode.py:582] Cuts processed until now is 12800.
234
+ 2025-11-17 17:45:58,858 INFO [streaming_decode.py:582] Cuts processed until now is 12900.
235
+ 2025-11-17 17:46:01,826 INFO [streaming_decode.py:582] Cuts processed until now is 13000.
236
+ 2025-11-17 17:46:04,052 INFO [streaming_decode.py:582] Cuts processed until now is 13100.
237
+ 2025-11-17 17:46:06,350 INFO [streaming_decode.py:582] Cuts processed until now is 13200.
238
+ 2025-11-17 17:46:08,745 INFO [streaming_decode.py:582] Cuts processed until now is 13300.
239
+ 2025-11-17 17:46:11,263 INFO [streaming_decode.py:582] Cuts processed until now is 13400.
240
+ 2025-11-17 17:46:13,637 INFO [streaming_decode.py:582] Cuts processed until now is 13500.
241
+ 2025-11-17 17:46:15,985 INFO [streaming_decode.py:582] Cuts processed until now is 13600.
242
+ 2025-11-17 17:46:18,490 INFO [streaming_decode.py:582] Cuts processed until now is 13700.
243
+ 2025-11-17 17:46:20,748 INFO [streaming_decode.py:582] Cuts processed until now is 13800.
244
+ 2025-11-17 17:46:23,131 INFO [streaming_decode.py:582] Cuts processed until now is 13900.
245
+ 2025-11-17 17:46:25,449 INFO [streaming_decode.py:582] Cuts processed until now is 14000.
246
+ 2025-11-17 17:46:27,783 INFO [streaming_decode.py:582] Cuts processed until now is 14100.
247
+ 2025-11-17 17:46:30,811 INFO [streaming_decode.py:582] Cuts processed until now is 14200.
248
+ 2025-11-17 17:46:33,157 INFO [streaming_decode.py:582] Cuts processed until now is 14300.
249
+ 2025-11-17 17:46:35,491 INFO [streaming_decode.py:582] Cuts processed until now is 14400.
250
+ 2025-11-17 17:46:37,800 INFO [streaming_decode.py:582] Cuts processed until now is 14500.
251
+ 2025-11-17 17:46:40,202 INFO [streaming_decode.py:582] Cuts processed until now is 14600.
252
+ 2025-11-17 17:46:43,145 INFO [streaming_decode.py:582] Cuts processed until now is 14700.
253
+ 2025-11-17 17:46:45,320 INFO [streaming_decode.py:582] Cuts processed until now is 14800.
254
+ 2025-11-17 17:46:47,530 INFO [streaming_decode.py:582] Cuts processed until now is 14900.
255
+ 2025-11-17 17:46:50,183 INFO [streaming_decode.py:582] Cuts processed until now is 15000.
256
+ 2025-11-17 17:46:52,441 INFO [streaming_decode.py:582] Cuts processed until now is 15100.
257
+ 2025-11-17 17:46:54,731 INFO [streaming_decode.py:582] Cuts processed until now is 15200.
258
+ 2025-11-17 17:46:56,948 INFO [streaming_decode.py:582] Cuts processed until now is 15300.
259
+ 2025-11-17 17:46:59,185 INFO [streaming_decode.py:582] Cuts processed until now is 15400.
260
+ 2025-11-17 17:47:01,389 INFO [streaming_decode.py:582] Cuts processed until now is 15500.
261
+ 2025-11-17 17:47:03,583 INFO [streaming_decode.py:582] Cuts processed until now is 15600.
262
+ 2025-11-17 17:47:05,925 INFO [streaming_decode.py:582] Cuts processed until now is 15700.
263
+ 2025-11-17 17:47:08,049 INFO [streaming_decode.py:582] Cuts processed until now is 15800.
264
+ 2025-11-17 17:47:20,220 INFO [streaming_decode.py:618] The transcripts are stored in tmp/exp-causal-80-epoch/streaming/fast_beam_search/recogs-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt
265
+ 2025-11-17 17:47:20,553 INFO [utils.py:670] [test-commonvoice-beam_4_max_contexts_4_max_states_32] %WER 5.57% [43361 / 778666, 3133 ins, 32018 del, 8210 sub ]
266
+ 2025-11-17 17:47:21,520 INFO [streaming_decode.py:627] Wrote detailed error stats to tmp/exp-causal-80-epoch/streaming/fast_beam_search/errs-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt
267
+ 2025-11-17 17:47:21,521 INFO [streaming_decode.py:641]
268
+ For test-commonvoice, WER of different settings are:
269
+ beam_4_max_contexts_4_max_states_32 5.57 best for test-commonvoice
270
+
271
+ 2025-11-17 17:47:21,528 INFO [streaming_decode.py:582] Cuts processed until now is 0.
272
+ 2025-11-17 17:47:22,565 INFO [streaming_decode.py:582] Cuts processed until now is 100.
273
+ 2025-11-17 17:47:23,489 INFO [streaming_decode.py:582] Cuts processed until now is 200.
274
+ 2025-11-17 17:47:24,176 INFO [streaming_decode.py:582] Cuts processed until now is 300.
275
+ 2025-11-17 17:47:24,971 INFO [streaming_decode.py:582] Cuts processed until now is 400.
276
+ 2025-11-17 17:47:25,853 INFO [streaming_decode.py:582] Cuts processed until now is 500.
277
+ 2025-11-17 17:47:26,795 INFO [streaming_decode.py:582] Cuts processed until now is 600.
278
+ 2025-11-17 17:47:27,624 INFO [streaming_decode.py:582] Cuts processed until now is 700.
279
+ 2025-11-17 17:47:28,393 INFO [streaming_decode.py:582] Cuts processed until now is 800.
280
+ 2025-11-17 17:47:29,289 INFO [streaming_decode.py:582] Cuts processed until now is 900.
281
+ 2025-11-17 17:47:44,309 INFO [streaming_decode.py:618] The transcripts are stored in tmp/exp-causal-80-epoch/streaming/fast_beam_search/recogs-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt
282
+ 2025-11-17 17:47:44,326 INFO [utils.py:670] [test-slr72-beam_4_max_contexts_4_max_states_32] %WER 2.18% [886 / 40600, 90 ins, 621 del, 175 sub ]
283
+ 2025-11-17 17:47:44,375 INFO [streaming_decode.py:627] Wrote detailed error stats to tmp/exp-causal-80-epoch/streaming/fast_beam_search/errs-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt
284
+ 2025-11-17 17:47:44,375 INFO [streaming_decode.py:641]
285
+ For test-slr72, WER of different settings are:
286
+ beam_4_max_contexts_4_max_states_32 2.18 best for test-slr72
287
+
288
+ 2025-11-17 17:47:44,375 INFO [streaming_decode.py:794] Done!
streaming/fast_beam_search/log-decode-epoch-80-avg-5-chunk-32-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model-2025-11-17-16-49-23 ADDED
@@ -0,0 +1,114 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2025-11-17 16:49:23,429 INFO [streaming_decode.py:677] Decoding started
2
+ 2025-11-17 16:49:23,429 INFO [streaming_decode.py:683] Device: cuda:0
3
+ 2025-11-17 16:49:23,430 INFO [lexicon.py:168] Loading pre-compiled data/lang_phone/Linv.pt
4
+ 2025-11-17 16:49:23,430 INFO [streaming_decode.py:691] {
5
+ "attention_decoder_attention_dim": 512,
6
+ "attention_decoder_dim": 512,
7
+ "attention_decoder_feedforward_dim": 2048,
8
+ "attention_decoder_num_heads": 8,
9
+ "attention_decoder_num_layers": 6,
10
+ "avg": 5,
11
+ "batch_idx_train": 0,
12
+ "beam": 4,
13
+ "best_train_epoch": -1,
14
+ "best_train_loss": Infinity,
15
+ "best_valid_epoch": -1,
16
+ "best_valid_loss": Infinity,
17
+ "blank_id": 0,
18
+ "bucketing_sampler": true,
19
+ "causal": true,
20
+ "chunk_size": "32",
21
+ "cnn_module_kernel": "31,31,15,15,15,31",
22
+ "concatenate_cuts": false,
23
+ "context_size": 2,
24
+ "decoder_dim": 512,
25
+ "decoding_method": "fast_beam_search",
26
+ "downsampling_factor": "1,2,4,8,4,2",
27
+ "drop_last": true,
28
+ "duration_factor": 1.0,
29
+ "enable_musan": true,
30
+ "enable_spec_aug": true,
31
+ "encoder_dim": "192,256,256,256,256,256",
32
+ "encoder_unmasked_dim": "192,192,192,192,192,192",
33
+ "env_info": {
34
+ "IP address": "127.0.1.1",
35
+ "hostname": "Bookbot-GPU1",
36
+ "icefall-git-branch": "master",
37
+ "icefall-git-date": "Tue Jan 7 16:20:44 2025",
38
+ "icefall-git-sha1": "77cd018f-dirty",
39
+ "icefall-path": "/mnt/Projects/Projects/ASR/icefall",
40
+ "k2-build-type": "Release",
41
+ "k2-git-date": "Thu Jul 25 03:46:03 2024",
42
+ "k2-git-sha1": "5735fa707f6091856d13ccd230aced6e9e64f815",
43
+ "k2-path": "/home/bookbot/miniconda3/envs/icefall/lib/python3.12/site-packages/k2/__init__.py",
44
+ "k2-version": "1.24.4",
45
+ "k2-with-cuda": true,
46
+ "lhotse-path": "/home/bookbot/miniconda3/envs/icefall/lib/python3.12/site-packages/lhotse/__init__.py",
47
+ "lhotse-version": "1.31.1.dev+git.4c43958.clean",
48
+ "python-version": "3.12",
49
+ "torch-cuda-available": true,
50
+ "torch-cuda-version": "12.4",
51
+ "torch-version": "2.4.0+cu124"
52
+ },
53
+ "epoch": 80,
54
+ "exp_dir": "tmp/exp-causal-80-epoch",
55
+ "feature_dim": 80,
56
+ "feedforward_dim": "512,768,768,768,768,768",
57
+ "gap": 1.0,
58
+ "ignore_id": -1,
59
+ "input_strategy": "PrecomputedFeatures",
60
+ "iter": 0,
61
+ "joiner_dim": 512,
62
+ "label_smoothing": 0.1,
63
+ "lang_dir": "data/lang_phone",
64
+ "left_context_frames": "128",
65
+ "log_interval": 50,
66
+ "manifest_dir": "data/fbank",
67
+ "max_contexts": 4,
68
+ "max_duration": 200.0,
69
+ "max_states": 32,
70
+ "num_active_paths": 4,
71
+ "num_buckets": 30,
72
+ "num_decode_streams": 1000,
73
+ "num_encoder_layers": "2,2,2,2,2,2",
74
+ "num_heads": "4,4,4,8,4,4",
75
+ "num_workers": 2,
76
+ "on_the_fly_feats": false,
77
+ "pos_dim": 48,
78
+ "pos_head_dim": "4",
79
+ "query_head_dim": "32",
80
+ "res_dir": "tmp/exp-causal-80-epoch/streaming/fast_beam_search",
81
+ "reset_interval": 200,
82
+ "return_cuts": true,
83
+ "shuffle": true,
84
+ "spec_aug_time_warp_factor": 80,
85
+ "subsampling_factor": 4,
86
+ "suffix": "epoch-80-avg-5-chunk-32-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model",
87
+ "unk_id": 1,
88
+ "use_attention_decoder": false,
89
+ "use_averaged_model": true,
90
+ "use_cr_ctc": false,
91
+ "use_ctc": false,
92
+ "use_transducer": true,
93
+ "valid_interval": 3000,
94
+ "value_head_dim": "12",
95
+ "vocab_size": 36,
96
+ "warm_step": 2000
97
+ }
98
+ 2025-11-17 16:49:23,430 INFO [streaming_decode.py:693] About to create model
99
+ 2025-11-17 16:49:23,627 INFO [streaming_decode.py:748] Calculating the averaged model over epoch range from 75 (excluded) to 80
100
+ 2025-11-17 16:49:24,973 INFO [streaming_decode.py:769] Number of model parameters: 22795007
101
+ 2025-11-17 16:49:24,973 INFO [multidataset.py:83] About to get Spanish Common Voice test cuts
102
+ 2025-11-17 16:49:24,973 INFO [multidataset.py:85] Loading Spanish Common Voice in lazy mode
103
+ 2025-11-17 16:49:24,974 INFO [multidataset.py:94] About to get SLR72 dataset test cuts
104
+ 2025-11-17 16:49:24,974 INFO [multidataset.py:96] Loading SLR72 dataset in lazy mode
105
+ 2025-11-17 16:49:25,102 INFO [streaming_decode.py:582] Cuts processed until now is 0.
106
+ 2025-11-17 16:49:27,727 INFO [streaming_decode.py:582] Cuts processed until now is 100.
107
+ 2025-11-17 16:49:29,936 INFO [streaming_decode.py:582] Cuts processed until now is 200.
108
+ 2025-11-17 16:49:31,730 INFO [streaming_decode.py:582] Cuts processed until now is 300.
109
+ 2025-11-17 16:49:34,212 INFO [streaming_decode.py:582] Cuts processed until now is 400.
110
+ 2025-11-17 16:49:36,336 INFO [streaming_decode.py:582] Cuts processed until now is 500.
111
+ 2025-11-17 16:49:38,919 INFO [streaming_decode.py:582] Cuts processed until now is 600.
112
+ 2025-11-17 16:49:41,052 INFO [streaming_decode.py:582] Cuts processed until now is 700.
113
+ 2025-11-17 16:49:43,304 INFO [streaming_decode.py:582] Cuts processed until now is 800.
114
+ 2025-11-17 16:49:45,956 INFO [streaming_decode.py:582] Cuts processed until now is 900.
streaming/fast_beam_search/log-decode-epoch-80-avg-5-chunk-32-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model-2025-11-17-16-58-25 ADDED
@@ -0,0 +1,288 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2025-11-17 16:58:25,918 INFO [streaming_decode.py:677] Decoding started
2
+ 2025-11-17 16:58:25,918 INFO [streaming_decode.py:683] Device: cuda:0
3
+ 2025-11-17 16:58:25,919 INFO [lexicon.py:168] Loading pre-compiled data/lang_phone/Linv.pt
4
+ 2025-11-17 16:58:25,919 INFO [streaming_decode.py:691] {
5
+ "attention_decoder_attention_dim": 512,
6
+ "attention_decoder_dim": 512,
7
+ "attention_decoder_feedforward_dim": 2048,
8
+ "attention_decoder_num_heads": 8,
9
+ "attention_decoder_num_layers": 6,
10
+ "avg": 5,
11
+ "batch_idx_train": 0,
12
+ "beam": 4,
13
+ "best_train_epoch": -1,
14
+ "best_train_loss": Infinity,
15
+ "best_valid_epoch": -1,
16
+ "best_valid_loss": Infinity,
17
+ "blank_id": 0,
18
+ "bucketing_sampler": true,
19
+ "causal": true,
20
+ "chunk_size": "32",
21
+ "cnn_module_kernel": "31,31,15,15,15,31",
22
+ "concatenate_cuts": false,
23
+ "context_size": 2,
24
+ "decoder_dim": 512,
25
+ "decoding_method": "fast_beam_search",
26
+ "downsampling_factor": "1,2,4,8,4,2",
27
+ "drop_last": true,
28
+ "duration_factor": 1.0,
29
+ "enable_musan": true,
30
+ "enable_spec_aug": true,
31
+ "encoder_dim": "192,256,256,256,256,256",
32
+ "encoder_unmasked_dim": "192,192,192,192,192,192",
33
+ "env_info": {
34
+ "IP address": "127.0.1.1",
35
+ "hostname": "Bookbot-GPU1",
36
+ "icefall-git-branch": "master",
37
+ "icefall-git-date": "Tue Jan 7 16:20:44 2025",
38
+ "icefall-git-sha1": "77cd018f-dirty",
39
+ "icefall-path": "/mnt/Projects/Projects/ASR/icefall",
40
+ "k2-build-type": "Release",
41
+ "k2-git-date": "Thu Jul 25 03:46:03 2024",
42
+ "k2-git-sha1": "5735fa707f6091856d13ccd230aced6e9e64f815",
43
+ "k2-path": "/home/bookbot/miniconda3/envs/icefall/lib/python3.12/site-packages/k2/__init__.py",
44
+ "k2-version": "1.24.4",
45
+ "k2-with-cuda": true,
46
+ "lhotse-path": "/home/bookbot/miniconda3/envs/icefall/lib/python3.12/site-packages/lhotse/__init__.py",
47
+ "lhotse-version": "1.31.1.dev+git.4c43958.clean",
48
+ "python-version": "3.12",
49
+ "torch-cuda-available": true,
50
+ "torch-cuda-version": "12.4",
51
+ "torch-version": "2.4.0+cu124"
52
+ },
53
+ "epoch": 80,
54
+ "exp_dir": "tmp/exp-causal-80-epoch",
55
+ "feature_dim": 80,
56
+ "feedforward_dim": "512,768,768,768,768,768",
57
+ "gap": 1.0,
58
+ "ignore_id": -1,
59
+ "input_strategy": "PrecomputedFeatures",
60
+ "iter": 0,
61
+ "joiner_dim": 512,
62
+ "label_smoothing": 0.1,
63
+ "lang_dir": "data/lang_phone",
64
+ "left_context_frames": "128",
65
+ "log_interval": 50,
66
+ "manifest_dir": "data/fbank",
67
+ "max_contexts": 4,
68
+ "max_duration": 200.0,
69
+ "max_states": 32,
70
+ "num_active_paths": 4,
71
+ "num_buckets": 30,
72
+ "num_decode_streams": 1000,
73
+ "num_encoder_layers": "2,2,2,2,2,2",
74
+ "num_heads": "4,4,4,8,4,4",
75
+ "num_workers": 2,
76
+ "on_the_fly_feats": false,
77
+ "pos_dim": 48,
78
+ "pos_head_dim": "4",
79
+ "query_head_dim": "32",
80
+ "res_dir": "tmp/exp-causal-80-epoch/streaming/fast_beam_search",
81
+ "reset_interval": 200,
82
+ "return_cuts": true,
83
+ "shuffle": true,
84
+ "spec_aug_time_warp_factor": 80,
85
+ "subsampling_factor": 4,
86
+ "suffix": "epoch-80-avg-5-chunk-32-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model",
87
+ "unk_id": 1,
88
+ "use_attention_decoder": false,
89
+ "use_averaged_model": true,
90
+ "use_cr_ctc": false,
91
+ "use_ctc": false,
92
+ "use_transducer": true,
93
+ "valid_interval": 3000,
94
+ "value_head_dim": "12",
95
+ "vocab_size": 36,
96
+ "warm_step": 2000
97
+ }
98
+ 2025-11-17 16:58:25,920 INFO [streaming_decode.py:693] About to create model
99
+ 2025-11-17 16:58:26,110 INFO [streaming_decode.py:748] Calculating the averaged model over epoch range from 75 (excluded) to 80
100
+ 2025-11-17 16:58:26,797 INFO [streaming_decode.py:769] Number of model parameters: 22795007
101
+ 2025-11-17 16:58:26,797 INFO [multidataset.py:83] About to get Spanish Common Voice test cuts
102
+ 2025-11-17 16:58:26,797 INFO [multidataset.py:85] Loading Spanish Common Voice in lazy mode
103
+ 2025-11-17 16:58:26,798 INFO [multidataset.py:94] About to get SLR72 dataset test cuts
104
+ 2025-11-17 16:58:26,798 INFO [multidataset.py:96] Loading SLR72 dataset in lazy mode
105
+ 2025-11-17 16:58:26,907 INFO [streaming_decode.py:582] Cuts processed until now is 0.
106
+ 2025-11-17 16:58:27,831 INFO [streaming_decode.py:582] Cuts processed until now is 100.
107
+ 2025-11-17 16:58:28,856 INFO [streaming_decode.py:582] Cuts processed until now is 200.
108
+ 2025-11-17 16:58:29,510 INFO [streaming_decode.py:582] Cuts processed until now is 300.
109
+ 2025-11-17 16:58:30,195 INFO [streaming_decode.py:582] Cuts processed until now is 400.
110
+ 2025-11-17 16:58:31,163 INFO [streaming_decode.py:582] Cuts processed until now is 500.
111
+ 2025-11-17 16:58:31,721 INFO [streaming_decode.py:582] Cuts processed until now is 600.
112
+ 2025-11-17 16:58:32,340 INFO [streaming_decode.py:582] Cuts processed until now is 700.
113
+ 2025-11-17 16:58:33,308 INFO [streaming_decode.py:582] Cuts processed until now is 800.
114
+ 2025-11-17 16:58:34,207 INFO [streaming_decode.py:582] Cuts processed until now is 900.
115
+ 2025-11-17 16:58:38,115 INFO [streaming_decode.py:582] Cuts processed until now is 1000.
116
+ 2025-11-17 16:58:41,358 INFO [streaming_decode.py:582] Cuts processed until now is 1100.
117
+ 2025-11-17 16:58:42,794 INFO [streaming_decode.py:582] Cuts processed until now is 1200.
118
+ 2025-11-17 16:58:44,425 INFO [streaming_decode.py:582] Cuts processed until now is 1300.
119
+ 2025-11-17 16:58:46,220 INFO [streaming_decode.py:582] Cuts processed until now is 1400.
120
+ 2025-11-17 16:58:46,966 INFO [streaming_decode.py:582] Cuts processed until now is 1500.
121
+ 2025-11-17 16:58:48,488 INFO [streaming_decode.py:582] Cuts processed until now is 1600.
122
+ 2025-11-17 16:58:50,482 INFO [streaming_decode.py:582] Cuts processed until now is 1700.
123
+ 2025-11-17 16:58:52,198 INFO [streaming_decode.py:582] Cuts processed until now is 1800.
124
+ 2025-11-17 16:58:53,732 INFO [streaming_decode.py:582] Cuts processed until now is 1900.
125
+ 2025-11-17 16:58:56,302 INFO [streaming_decode.py:582] Cuts processed until now is 2000.
126
+ 2025-11-17 16:58:58,125 INFO [streaming_decode.py:582] Cuts processed until now is 2100.
127
+ 2025-11-17 16:59:00,007 INFO [streaming_decode.py:582] Cuts processed until now is 2200.
128
+ 2025-11-17 16:59:01,803 INFO [streaming_decode.py:582] Cuts processed until now is 2300.
129
+ 2025-11-17 16:59:03,350 INFO [streaming_decode.py:582] Cuts processed until now is 2400.
130
+ 2025-11-17 16:59:05,161 INFO [streaming_decode.py:582] Cuts processed until now is 2500.
131
+ 2025-11-17 16:59:06,958 INFO [streaming_decode.py:582] Cuts processed until now is 2600.
132
+ 2025-11-17 16:59:08,492 INFO [streaming_decode.py:582] Cuts processed until now is 2700.
133
+ 2025-11-17 16:59:10,030 INFO [streaming_decode.py:582] Cuts processed until now is 2800.
134
+ 2025-11-17 16:59:11,449 INFO [streaming_decode.py:582] Cuts processed until now is 2900.
135
+ 2025-11-17 16:59:12,856 INFO [streaming_decode.py:582] Cuts processed until now is 3000.
136
+ 2025-11-17 16:59:15,102 INFO [streaming_decode.py:582] Cuts processed until now is 3100.
137
+ 2025-11-17 16:59:16,632 INFO [streaming_decode.py:582] Cuts processed until now is 3200.
138
+ 2025-11-17 16:59:18,294 INFO [streaming_decode.py:582] Cuts processed until now is 3300.
139
+ 2025-11-17 16:59:19,897 INFO [streaming_decode.py:582] Cuts processed until now is 3400.
140
+ 2025-11-17 16:59:21,579 INFO [streaming_decode.py:582] Cuts processed until now is 3500.
141
+ 2025-11-17 16:59:23,123 INFO [streaming_decode.py:582] Cuts processed until now is 3600.
142
+ 2025-11-17 16:59:24,814 INFO [streaming_decode.py:582] Cuts processed until now is 3700.
143
+ 2025-11-17 16:59:26,474 INFO [streaming_decode.py:582] Cuts processed until now is 3800.
144
+ 2025-11-17 16:59:28,226 INFO [streaming_decode.py:582] Cuts processed until now is 3900.
145
+ 2025-11-17 16:59:29,871 INFO [streaming_decode.py:582] Cuts processed until now is 4000.
146
+ 2025-11-17 16:59:31,481 INFO [streaming_decode.py:582] Cuts processed until now is 4100.
147
+ 2025-11-17 16:59:33,226 INFO [streaming_decode.py:582] Cuts processed until now is 4200.
148
+ 2025-11-17 16:59:35,015 INFO [streaming_decode.py:582] Cuts processed until now is 4300.
149
+ 2025-11-17 16:59:36,756 INFO [streaming_decode.py:582] Cuts processed until now is 4400.
150
+ 2025-11-17 16:59:38,306 INFO [streaming_decode.py:582] Cuts processed until now is 4500.
151
+ 2025-11-17 16:59:39,664 INFO [streaming_decode.py:582] Cuts processed until now is 4600.
152
+ 2025-11-17 16:59:41,482 INFO [streaming_decode.py:582] Cuts processed until now is 4700.
153
+ 2025-11-17 16:59:44,271 INFO [streaming_decode.py:582] Cuts processed until now is 4800.
154
+ 2025-11-17 16:59:45,803 INFO [streaming_decode.py:582] Cuts processed until now is 4900.
155
+ 2025-11-17 16:59:47,689 INFO [streaming_decode.py:582] Cuts processed until now is 5000.
156
+ 2025-11-17 16:59:49,508 INFO [streaming_decode.py:582] Cuts processed until now is 5100.
157
+ 2025-11-17 16:59:51,421 INFO [streaming_decode.py:582] Cuts processed until now is 5200.
158
+ 2025-11-17 16:59:53,016 INFO [streaming_decode.py:582] Cuts processed until now is 5300.
159
+ 2025-11-17 16:59:54,782 INFO [streaming_decode.py:582] Cuts processed until now is 5400.
160
+ 2025-11-17 16:59:56,582 INFO [streaming_decode.py:582] Cuts processed until now is 5500.
161
+ 2025-11-17 16:59:58,361 INFO [streaming_decode.py:582] Cuts processed until now is 5600.
162
+ 2025-11-17 17:00:00,234 INFO [streaming_decode.py:582] Cuts processed until now is 5700.
163
+ 2025-11-17 17:00:01,916 INFO [streaming_decode.py:582] Cuts processed until now is 5800.
164
+ 2025-11-17 17:00:04,696 INFO [streaming_decode.py:582] Cuts processed until now is 5900.
165
+ 2025-11-17 17:00:06,286 INFO [streaming_decode.py:582] Cuts processed until now is 6000.
166
+ 2025-11-17 17:00:07,719 INFO [streaming_decode.py:582] Cuts processed until now is 6100.
167
+ 2025-11-17 17:00:09,408 INFO [streaming_decode.py:582] Cuts processed until now is 6200.
168
+ 2025-11-17 17:00:11,195 INFO [streaming_decode.py:582] Cuts processed until now is 6300.
169
+ 2025-11-17 17:00:12,789 INFO [streaming_decode.py:582] Cuts processed until now is 6400.
170
+ 2025-11-17 17:00:14,575 INFO [streaming_decode.py:582] Cuts processed until now is 6500.
171
+ 2025-11-17 17:00:16,293 INFO [streaming_decode.py:582] Cuts processed until now is 6600.
172
+ 2025-11-17 17:00:17,730 INFO [streaming_decode.py:582] Cuts processed until now is 6700.
173
+ 2025-11-17 17:00:19,350 INFO [streaming_decode.py:582] Cuts processed until now is 6800.
174
+ 2025-11-17 17:00:20,978 INFO [streaming_decode.py:582] Cuts processed until now is 6900.
175
+ 2025-11-17 17:00:22,767 INFO [streaming_decode.py:582] Cuts processed until now is 7000.
176
+ 2025-11-17 17:00:24,289 INFO [streaming_decode.py:582] Cuts processed until now is 7100.
177
+ 2025-11-17 17:00:26,735 INFO [streaming_decode.py:582] Cuts processed until now is 7200.
178
+ 2025-11-17 17:00:28,347 INFO [streaming_decode.py:582] Cuts processed until now is 7300.
179
+ 2025-11-17 17:00:30,283 INFO [streaming_decode.py:582] Cuts processed until now is 7400.
180
+ 2025-11-17 17:00:32,034 INFO [streaming_decode.py:582] Cuts processed until now is 7500.
181
+ 2025-11-17 17:00:33,643 INFO [streaming_decode.py:582] Cuts processed until now is 7600.
182
+ 2025-11-17 17:00:35,104 INFO [streaming_decode.py:582] Cuts processed until now is 7700.
183
+ 2025-11-17 17:00:37,214 INFO [streaming_decode.py:582] Cuts processed until now is 7800.
184
+ 2025-11-17 17:00:39,024 INFO [streaming_decode.py:582] Cuts processed until now is 7900.
185
+ 2025-11-17 17:00:40,713 INFO [streaming_decode.py:582] Cuts processed until now is 8000.
186
+ 2025-11-17 17:00:42,193 INFO [streaming_decode.py:582] Cuts processed until now is 8100.
187
+ 2025-11-17 17:00:44,074 INFO [streaming_decode.py:582] Cuts processed until now is 8200.
188
+ 2025-11-17 17:00:45,860 INFO [streaming_decode.py:582] Cuts processed until now is 8300.
189
+ 2025-11-17 17:00:47,738 INFO [streaming_decode.py:582] Cuts processed until now is 8400.
190
+ 2025-11-17 17:00:49,260 INFO [streaming_decode.py:582] Cuts processed until now is 8500.
191
+ 2025-11-17 17:00:50,999 INFO [streaming_decode.py:582] Cuts processed until now is 8600.
192
+ 2025-11-17 17:00:52,649 INFO [streaming_decode.py:582] Cuts processed until now is 8700.
193
+ 2025-11-17 17:00:55,324 INFO [streaming_decode.py:582] Cuts processed until now is 8800.
194
+ 2025-11-17 17:00:57,170 INFO [streaming_decode.py:582] Cuts processed until now is 8900.
195
+ 2025-11-17 17:00:58,886 INFO [streaming_decode.py:582] Cuts processed until now is 9000.
196
+ 2025-11-17 17:01:00,493 INFO [streaming_decode.py:582] Cuts processed until now is 9100.
197
+ 2025-11-17 17:01:02,415 INFO [streaming_decode.py:582] Cuts processed until now is 9200.
198
+ 2025-11-17 17:01:04,151 INFO [streaming_decode.py:582] Cuts processed until now is 9300.
199
+ 2025-11-17 17:01:05,942 INFO [streaming_decode.py:582] Cuts processed until now is 9400.
200
+ 2025-11-17 17:01:07,707 INFO [streaming_decode.py:582] Cuts processed until now is 9500.
201
+ 2025-11-17 17:01:09,332 INFO [streaming_decode.py:582] Cuts processed until now is 9600.
202
+ 2025-11-17 17:01:10,824 INFO [streaming_decode.py:582] Cuts processed until now is 9700.
203
+ 2025-11-17 17:01:12,306 INFO [streaming_decode.py:582] Cuts processed until now is 9800.
204
+ 2025-11-17 17:01:14,041 INFO [streaming_decode.py:582] Cuts processed until now is 9900.
205
+ 2025-11-17 17:01:16,023 INFO [streaming_decode.py:582] Cuts processed until now is 10000.
206
+ 2025-11-17 17:01:17,811 INFO [streaming_decode.py:582] Cuts processed until now is 10100.
207
+ 2025-11-17 17:01:19,837 INFO [streaming_decode.py:582] Cuts processed until now is 10200.
208
+ 2025-11-17 17:01:22,729 INFO [streaming_decode.py:582] Cuts processed until now is 10300.
209
+ 2025-11-17 17:01:24,488 INFO [streaming_decode.py:582] Cuts processed until now is 10400.
210
+ 2025-11-17 17:01:26,280 INFO [streaming_decode.py:582] Cuts processed until now is 10500.
211
+ 2025-11-17 17:01:28,121 INFO [streaming_decode.py:582] Cuts processed until now is 10600.
212
+ 2025-11-17 17:01:29,986 INFO [streaming_decode.py:582] Cuts processed until now is 10700.
213
+ 2025-11-17 17:01:31,736 INFO [streaming_decode.py:582] Cuts processed until now is 10800.
214
+ 2025-11-17 17:01:33,523 INFO [streaming_decode.py:582] Cuts processed until now is 10900.
215
+ 2025-11-17 17:01:35,154 INFO [streaming_decode.py:582] Cuts processed until now is 11000.
216
+ 2025-11-17 17:01:36,634 INFO [streaming_decode.py:582] Cuts processed until now is 11100.
217
+ 2025-11-17 17:01:38,295 INFO [streaming_decode.py:582] Cuts processed until now is 11200.
218
+ 2025-11-17 17:01:40,087 INFO [streaming_decode.py:582] Cuts processed until now is 11300.
219
+ 2025-11-17 17:01:41,948 INFO [streaming_decode.py:582] Cuts processed until now is 11400.
220
+ 2025-11-17 17:01:44,533 INFO [streaming_decode.py:582] Cuts processed until now is 11500.
221
+ 2025-11-17 17:01:46,436 INFO [streaming_decode.py:582] Cuts processed until now is 11600.
222
+ 2025-11-17 17:01:48,217 INFO [streaming_decode.py:582] Cuts processed until now is 11700.
223
+ 2025-11-17 17:01:49,826 INFO [streaming_decode.py:582] Cuts processed until now is 11800.
224
+ 2025-11-17 17:01:51,413 INFO [streaming_decode.py:582] Cuts processed until now is 11900.
225
+ 2025-11-17 17:01:53,127 INFO [streaming_decode.py:582] Cuts processed until now is 12000.
226
+ 2025-11-17 17:01:54,865 INFO [streaming_decode.py:582] Cuts processed until now is 12100.
227
+ 2025-11-17 17:01:56,805 INFO [streaming_decode.py:582] Cuts processed until now is 12200.
228
+ 2025-11-17 17:01:58,441 INFO [streaming_decode.py:582] Cuts processed until now is 12300.
229
+ 2025-11-17 17:02:00,295 INFO [streaming_decode.py:582] Cuts processed until now is 12400.
230
+ 2025-11-17 17:02:02,218 INFO [streaming_decode.py:582] Cuts processed until now is 12500.
231
+ 2025-11-17 17:02:03,883 INFO [streaming_decode.py:582] Cuts processed until now is 12600.
232
+ 2025-11-17 17:02:05,785 INFO [streaming_decode.py:582] Cuts processed until now is 12700.
233
+ 2025-11-17 17:02:07,844 INFO [streaming_decode.py:582] Cuts processed until now is 12800.
234
+ 2025-11-17 17:02:09,766 INFO [streaming_decode.py:582] Cuts processed until now is 12900.
235
+ 2025-11-17 17:02:11,628 INFO [streaming_decode.py:582] Cuts processed until now is 13000.
236
+ 2025-11-17 17:02:13,238 INFO [streaming_decode.py:582] Cuts processed until now is 13100.
237
+ 2025-11-17 17:02:16,110 INFO [streaming_decode.py:582] Cuts processed until now is 13200.
238
+ 2025-11-17 17:02:18,085 INFO [streaming_decode.py:582] Cuts processed until now is 13300.
239
+ 2025-11-17 17:02:19,800 INFO [streaming_decode.py:582] Cuts processed until now is 13400.
240
+ 2025-11-17 17:02:21,221 INFO [streaming_decode.py:582] Cuts processed until now is 13500.
241
+ 2025-11-17 17:02:22,894 INFO [streaming_decode.py:582] Cuts processed until now is 13600.
242
+ 2025-11-17 17:02:24,457 INFO [streaming_decode.py:582] Cuts processed until now is 13700.
243
+ 2025-11-17 17:02:25,883 INFO [streaming_decode.py:582] Cuts processed until now is 13800.
244
+ 2025-11-17 17:02:27,362 INFO [streaming_decode.py:582] Cuts processed until now is 13900.
245
+ 2025-11-17 17:02:28,842 INFO [streaming_decode.py:582] Cuts processed until now is 14000.
246
+ 2025-11-17 17:02:30,517 INFO [streaming_decode.py:582] Cuts processed until now is 14100.
247
+ 2025-11-17 17:02:32,591 INFO [streaming_decode.py:582] Cuts processed until now is 14200.
248
+ 2025-11-17 17:02:34,284 INFO [streaming_decode.py:582] Cuts processed until now is 14300.
249
+ 2025-11-17 17:02:36,641 INFO [streaming_decode.py:582] Cuts processed until now is 14400.
250
+ 2025-11-17 17:02:38,550 INFO [streaming_decode.py:582] Cuts processed until now is 14500.
251
+ 2025-11-17 17:02:40,272 INFO [streaming_decode.py:582] Cuts processed until now is 14600.
252
+ 2025-11-17 17:02:41,849 INFO [streaming_decode.py:582] Cuts processed until now is 14700.
253
+ 2025-11-17 17:02:43,812 INFO [streaming_decode.py:582] Cuts processed until now is 14800.
254
+ 2025-11-17 17:02:45,478 INFO [streaming_decode.py:582] Cuts processed until now is 14900.
255
+ 2025-11-17 17:02:47,373 INFO [streaming_decode.py:582] Cuts processed until now is 15000.
256
+ 2025-11-17 17:02:49,236 INFO [streaming_decode.py:582] Cuts processed until now is 15100.
257
+ 2025-11-17 17:02:50,882 INFO [streaming_decode.py:582] Cuts processed until now is 15200.
258
+ 2025-11-17 17:02:52,659 INFO [streaming_decode.py:582] Cuts processed until now is 15300.
259
+ 2025-11-17 17:02:54,397 INFO [streaming_decode.py:582] Cuts processed until now is 15400.
260
+ 2025-11-17 17:02:57,086 INFO [streaming_decode.py:582] Cuts processed until now is 15500.
261
+ 2025-11-17 17:02:57,933 INFO [streaming_decode.py:582] Cuts processed until now is 15600.
262
+ 2025-11-17 17:02:59,762 INFO [streaming_decode.py:582] Cuts processed until now is 15700.
263
+ 2025-11-17 17:03:01,696 INFO [streaming_decode.py:582] Cuts processed until now is 15800.
264
+ 2025-11-17 17:03:10,884 INFO [streaming_decode.py:618] The transcripts are stored in tmp/exp-causal-80-epoch/streaming/fast_beam_search/recogs-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt
265
+ 2025-11-17 17:03:11,198 INFO [utils.py:670] [test-commonvoice-beam_4_max_contexts_4_max_states_32] %WER 6.39% [49746 / 778666, 2817 ins, 39514 del, 7415 sub ]
266
+ 2025-11-17 17:03:12,033 INFO [streaming_decode.py:627] Wrote detailed error stats to tmp/exp-causal-80-epoch/streaming/fast_beam_search/errs-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt
267
+ 2025-11-17 17:03:12,033 INFO [streaming_decode.py:641]
268
+ For test-commonvoice, WER of different settings are:
269
+ beam_4_max_contexts_4_max_states_32 6.39 best for test-commonvoice
270
+
271
+ 2025-11-17 17:03:12,039 INFO [streaming_decode.py:582] Cuts processed until now is 0.
272
+ 2025-11-17 17:03:12,740 INFO [streaming_decode.py:582] Cuts processed until now is 100.
273
+ 2025-11-17 17:03:13,574 INFO [streaming_decode.py:582] Cuts processed until now is 200.
274
+ 2025-11-17 17:03:14,639 INFO [streaming_decode.py:582] Cuts processed until now is 300.
275
+ 2025-11-17 17:03:15,667 INFO [streaming_decode.py:582] Cuts processed until now is 400.
276
+ 2025-11-17 17:03:16,800 INFO [streaming_decode.py:582] Cuts processed until now is 500.
277
+ 2025-11-17 17:03:17,710 INFO [streaming_decode.py:582] Cuts processed until now is 600.
278
+ 2025-11-17 17:03:18,546 INFO [streaming_decode.py:582] Cuts processed until now is 700.
279
+ 2025-11-17 17:03:19,572 INFO [streaming_decode.py:582] Cuts processed until now is 800.
280
+ 2025-11-17 17:03:20,341 INFO [streaming_decode.py:582] Cuts processed until now is 900.
281
+ 2025-11-17 17:03:29,501 INFO [streaming_decode.py:618] The transcripts are stored in tmp/exp-causal-80-epoch/streaming/fast_beam_search/recogs-test-slr72-epoch-80-avg-5-chunk-32-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt
282
+ 2025-11-17 17:03:29,521 INFO [utils.py:670] [test-slr72-beam_4_max_contexts_4_max_states_32] %WER 2.44% [989 / 40600, 73 ins, 777 del, 139 sub ]
283
+ 2025-11-17 17:03:29,573 INFO [streaming_decode.py:627] Wrote detailed error stats to tmp/exp-causal-80-epoch/streaming/fast_beam_search/errs-test-slr72-epoch-80-avg-5-chunk-32-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt
284
+ 2025-11-17 17:03:29,573 INFO [streaming_decode.py:641]
285
+ For test-slr72, WER of different settings are:
286
+ beam_4_max_contexts_4_max_states_32 2.44 best for test-slr72
287
+
288
+ 2025-11-17 17:03:29,573 INFO [streaming_decode.py:794] Done!
streaming/fast_beam_search/recogs-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/fast_beam_search/recogs-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/fast_beam_search/recogs-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/fast_beam_search/recogs-test-slr72-epoch-80-avg-5-chunk-32-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/fast_beam_search/wer-summary-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ settings WER
2
+ beam_4_max_contexts_4_max_states_32 5.57
streaming/fast_beam_search/wer-summary-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ settings WER
2
+ beam_4_max_contexts_4_max_states_32 6.39
streaming/fast_beam_search/wer-summary-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ settings WER
2
+ beam_4_max_contexts_4_max_states_32 2.18
streaming/fast_beam_search/wer-summary-test-slr72-epoch-80-avg-5-chunk-32-left-context-128-beam-4-max-contexts-4-max-states-32-use-averaged-model.txt ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ settings WER
2
+ beam_4_max_contexts_4_max_states_32 2.44
streaming/greedy_search/errs-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/greedy_search/errs-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/greedy_search/errs-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/greedy_search/errs-test-slr72-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/greedy_search/log-decode-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model-2025-11-17-17-37-14 ADDED
@@ -0,0 +1,288 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2025-11-17 17:37:14,887 INFO [streaming_decode.py:677] Decoding started
2
+ 2025-11-17 17:37:14,887 INFO [streaming_decode.py:683] Device: cuda:0
3
+ 2025-11-17 17:37:14,887 INFO [lexicon.py:168] Loading pre-compiled data/lang_phone/Linv.pt
4
+ 2025-11-17 17:37:14,888 INFO [streaming_decode.py:691] {
5
+ "attention_decoder_attention_dim": 512,
6
+ "attention_decoder_dim": 512,
7
+ "attention_decoder_feedforward_dim": 2048,
8
+ "attention_decoder_num_heads": 8,
9
+ "attention_decoder_num_layers": 6,
10
+ "avg": 5,
11
+ "batch_idx_train": 0,
12
+ "beam": 4,
13
+ "best_train_epoch": -1,
14
+ "best_train_loss": Infinity,
15
+ "best_valid_epoch": -1,
16
+ "best_valid_loss": Infinity,
17
+ "blank_id": 0,
18
+ "bucketing_sampler": true,
19
+ "causal": true,
20
+ "chunk_size": "16",
21
+ "cnn_module_kernel": "31,31,15,15,15,31",
22
+ "concatenate_cuts": false,
23
+ "context_size": 2,
24
+ "decoder_dim": 512,
25
+ "decoding_method": "greedy_search",
26
+ "downsampling_factor": "1,2,4,8,4,2",
27
+ "drop_last": true,
28
+ "duration_factor": 1.0,
29
+ "enable_musan": true,
30
+ "enable_spec_aug": true,
31
+ "encoder_dim": "192,256,256,256,256,256",
32
+ "encoder_unmasked_dim": "192,192,192,192,192,192",
33
+ "env_info": {
34
+ "IP address": "127.0.1.1",
35
+ "hostname": "Bookbot-GPU1",
36
+ "icefall-git-branch": "master",
37
+ "icefall-git-date": "Tue Jan 7 16:20:44 2025",
38
+ "icefall-git-sha1": "77cd018f-dirty",
39
+ "icefall-path": "/mnt/Projects/Projects/ASR/icefall",
40
+ "k2-build-type": "Release",
41
+ "k2-git-date": "Thu Jul 25 03:46:03 2024",
42
+ "k2-git-sha1": "5735fa707f6091856d13ccd230aced6e9e64f815",
43
+ "k2-path": "/home/bookbot/miniconda3/envs/icefall/lib/python3.12/site-packages/k2/__init__.py",
44
+ "k2-version": "1.24.4",
45
+ "k2-with-cuda": true,
46
+ "lhotse-path": "/home/bookbot/miniconda3/envs/icefall/lib/python3.12/site-packages/lhotse/__init__.py",
47
+ "lhotse-version": "1.31.1.dev+git.4c43958.clean",
48
+ "python-version": "3.12",
49
+ "torch-cuda-available": true,
50
+ "torch-cuda-version": "12.4",
51
+ "torch-version": "2.4.0+cu124"
52
+ },
53
+ "epoch": 80,
54
+ "exp_dir": "tmp/exp-causal-80-epoch",
55
+ "feature_dim": 80,
56
+ "feedforward_dim": "512,768,768,768,768,768",
57
+ "gap": 1.0,
58
+ "ignore_id": -1,
59
+ "input_strategy": "PrecomputedFeatures",
60
+ "iter": 0,
61
+ "joiner_dim": 512,
62
+ "label_smoothing": 0.1,
63
+ "lang_dir": "data/lang_phone",
64
+ "left_context_frames": "128",
65
+ "log_interval": 50,
66
+ "manifest_dir": "data/fbank",
67
+ "max_contexts": 4,
68
+ "max_duration": 200.0,
69
+ "max_states": 32,
70
+ "num_active_paths": 4,
71
+ "num_buckets": 30,
72
+ "num_decode_streams": 1000,
73
+ "num_encoder_layers": "2,2,2,2,2,2",
74
+ "num_heads": "4,4,4,8,4,4",
75
+ "num_workers": 2,
76
+ "on_the_fly_feats": false,
77
+ "pos_dim": 48,
78
+ "pos_head_dim": "4",
79
+ "query_head_dim": "32",
80
+ "res_dir": "tmp/exp-causal-80-epoch/streaming/greedy_search",
81
+ "reset_interval": 200,
82
+ "return_cuts": true,
83
+ "shuffle": true,
84
+ "spec_aug_time_warp_factor": 80,
85
+ "subsampling_factor": 4,
86
+ "suffix": "epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model",
87
+ "unk_id": 1,
88
+ "use_attention_decoder": false,
89
+ "use_averaged_model": true,
90
+ "use_cr_ctc": false,
91
+ "use_ctc": false,
92
+ "use_transducer": true,
93
+ "valid_interval": 3000,
94
+ "value_head_dim": "12",
95
+ "vocab_size": 36,
96
+ "warm_step": 2000
97
+ }
98
+ 2025-11-17 17:37:14,889 INFO [streaming_decode.py:693] About to create model
99
+ 2025-11-17 17:37:15,078 INFO [streaming_decode.py:748] Calculating the averaged model over epoch range from 75 (excluded) to 80
100
+ 2025-11-17 17:37:15,818 INFO [streaming_decode.py:769] Number of model parameters: 22795007
101
+ 2025-11-17 17:37:15,818 INFO [multidataset.py:83] About to get Spanish Common Voice test cuts
102
+ 2025-11-17 17:37:15,818 INFO [multidataset.py:85] Loading Spanish Common Voice in lazy mode
103
+ 2025-11-17 17:37:15,819 INFO [multidataset.py:94] About to get SLR72 dataset test cuts
104
+ 2025-11-17 17:37:15,819 INFO [multidataset.py:96] Loading SLR72 dataset in lazy mode
105
+ 2025-11-17 17:37:15,930 INFO [streaming_decode.py:582] Cuts processed until now is 0.
106
+ 2025-11-17 17:37:16,736 INFO [streaming_decode.py:582] Cuts processed until now is 100.
107
+ 2025-11-17 17:37:17,694 INFO [streaming_decode.py:582] Cuts processed until now is 200.
108
+ 2025-11-17 17:37:18,477 INFO [streaming_decode.py:582] Cuts processed until now is 300.
109
+ 2025-11-17 17:37:19,198 INFO [streaming_decode.py:582] Cuts processed until now is 400.
110
+ 2025-11-17 17:37:20,186 INFO [streaming_decode.py:582] Cuts processed until now is 500.
111
+ 2025-11-17 17:37:20,871 INFO [streaming_decode.py:582] Cuts processed until now is 600.
112
+ 2025-11-17 17:37:21,703 INFO [streaming_decode.py:582] Cuts processed until now is 700.
113
+ 2025-11-17 17:37:22,470 INFO [streaming_decode.py:582] Cuts processed until now is 800.
114
+ 2025-11-17 17:37:23,172 INFO [streaming_decode.py:582] Cuts processed until now is 900.
115
+ 2025-11-17 17:37:25,983 INFO [streaming_decode.py:582] Cuts processed until now is 1000.
116
+ 2025-11-17 17:37:28,242 INFO [streaming_decode.py:582] Cuts processed until now is 1100.
117
+ 2025-11-17 17:37:29,500 INFO [streaming_decode.py:582] Cuts processed until now is 1200.
118
+ 2025-11-17 17:37:30,832 INFO [streaming_decode.py:582] Cuts processed until now is 1300.
119
+ 2025-11-17 17:37:31,931 INFO [streaming_decode.py:582] Cuts processed until now is 1400.
120
+ 2025-11-17 17:37:32,919 INFO [streaming_decode.py:582] Cuts processed until now is 1500.
121
+ 2025-11-17 17:37:34,276 INFO [streaming_decode.py:582] Cuts processed until now is 1600.
122
+ 2025-11-17 17:37:35,428 INFO [streaming_decode.py:582] Cuts processed until now is 1700.
123
+ 2025-11-17 17:37:36,740 INFO [streaming_decode.py:582] Cuts processed until now is 1800.
124
+ 2025-11-17 17:37:38,103 INFO [streaming_decode.py:582] Cuts processed until now is 1900.
125
+ 2025-11-17 17:37:39,811 INFO [streaming_decode.py:582] Cuts processed until now is 2000.
126
+ 2025-11-17 17:37:41,372 INFO [streaming_decode.py:582] Cuts processed until now is 2100.
127
+ 2025-11-17 17:37:42,540 INFO [streaming_decode.py:582] Cuts processed until now is 2200.
128
+ 2025-11-17 17:37:44,044 INFO [streaming_decode.py:582] Cuts processed until now is 2300.
129
+ 2025-11-17 17:37:45,270 INFO [streaming_decode.py:582] Cuts processed until now is 2400.
130
+ 2025-11-17 17:37:46,564 INFO [streaming_decode.py:582] Cuts processed until now is 2500.
131
+ 2025-11-17 17:37:48,008 INFO [streaming_decode.py:582] Cuts processed until now is 2600.
132
+ 2025-11-17 17:37:49,525 INFO [streaming_decode.py:582] Cuts processed until now is 2700.
133
+ 2025-11-17 17:37:50,956 INFO [streaming_decode.py:582] Cuts processed until now is 2800.
134
+ 2025-11-17 17:37:52,336 INFO [streaming_decode.py:582] Cuts processed until now is 2900.
135
+ 2025-11-17 17:37:53,854 INFO [streaming_decode.py:582] Cuts processed until now is 3000.
136
+ 2025-11-17 17:37:55,511 INFO [streaming_decode.py:582] Cuts processed until now is 3100.
137
+ 2025-11-17 17:37:56,972 INFO [streaming_decode.py:582] Cuts processed until now is 3200.
138
+ 2025-11-17 17:37:58,389 INFO [streaming_decode.py:582] Cuts processed until now is 3300.
139
+ 2025-11-17 17:38:00,021 INFO [streaming_decode.py:582] Cuts processed until now is 3400.
140
+ 2025-11-17 17:38:01,367 INFO [streaming_decode.py:582] Cuts processed until now is 3500.
141
+ 2025-11-17 17:38:02,788 INFO [streaming_decode.py:582] Cuts processed until now is 3600.
142
+ 2025-11-17 17:38:04,069 INFO [streaming_decode.py:582] Cuts processed until now is 3700.
143
+ 2025-11-17 17:38:05,386 INFO [streaming_decode.py:582] Cuts processed until now is 3800.
144
+ 2025-11-17 17:38:06,945 INFO [streaming_decode.py:582] Cuts processed until now is 3900.
145
+ 2025-11-17 17:38:08,319 INFO [streaming_decode.py:582] Cuts processed until now is 4000.
146
+ 2025-11-17 17:38:09,250 INFO [streaming_decode.py:582] Cuts processed until now is 4100.
147
+ 2025-11-17 17:38:10,759 INFO [streaming_decode.py:582] Cuts processed until now is 4200.
148
+ 2025-11-17 17:38:12,109 INFO [streaming_decode.py:582] Cuts processed until now is 4300.
149
+ 2025-11-17 17:38:13,525 INFO [streaming_decode.py:582] Cuts processed until now is 4400.
150
+ 2025-11-17 17:38:14,838 INFO [streaming_decode.py:582] Cuts processed until now is 4500.
151
+ 2025-11-17 17:38:16,009 INFO [streaming_decode.py:582] Cuts processed until now is 4600.
152
+ 2025-11-17 17:38:17,404 INFO [streaming_decode.py:582] Cuts processed until now is 4700.
153
+ 2025-11-17 17:38:18,823 INFO [streaming_decode.py:582] Cuts processed until now is 4800.
154
+ 2025-11-17 17:38:20,037 INFO [streaming_decode.py:582] Cuts processed until now is 4900.
155
+ 2025-11-17 17:38:21,380 INFO [streaming_decode.py:582] Cuts processed until now is 5000.
156
+ 2025-11-17 17:38:22,755 INFO [streaming_decode.py:582] Cuts processed until now is 5100.
157
+ 2025-11-17 17:38:24,320 INFO [streaming_decode.py:582] Cuts processed until now is 5200.
158
+ 2025-11-17 17:38:25,820 INFO [streaming_decode.py:582] Cuts processed until now is 5300.
159
+ 2025-11-17 17:38:27,059 INFO [streaming_decode.py:582] Cuts processed until now is 5400.
160
+ 2025-11-17 17:38:28,575 INFO [streaming_decode.py:582] Cuts processed until now is 5500.
161
+ 2025-11-17 17:38:29,702 INFO [streaming_decode.py:582] Cuts processed until now is 5600.
162
+ 2025-11-17 17:38:31,306 INFO [streaming_decode.py:582] Cuts processed until now is 5700.
163
+ 2025-11-17 17:38:32,618 INFO [streaming_decode.py:582] Cuts processed until now is 5800.
164
+ 2025-11-17 17:38:33,995 INFO [streaming_decode.py:582] Cuts processed until now is 5900.
165
+ 2025-11-17 17:38:35,376 INFO [streaming_decode.py:582] Cuts processed until now is 6000.
166
+ 2025-11-17 17:38:36,914 INFO [streaming_decode.py:582] Cuts processed until now is 6100.
167
+ 2025-11-17 17:38:38,493 INFO [streaming_decode.py:582] Cuts processed until now is 6200.
168
+ 2025-11-17 17:38:39,943 INFO [streaming_decode.py:582] Cuts processed until now is 6300.
169
+ 2025-11-17 17:38:41,181 INFO [streaming_decode.py:582] Cuts processed until now is 6400.
170
+ 2025-11-17 17:38:42,640 INFO [streaming_decode.py:582] Cuts processed until now is 6500.
171
+ 2025-11-17 17:38:43,889 INFO [streaming_decode.py:582] Cuts processed until now is 6600.
172
+ 2025-11-17 17:38:45,068 INFO [streaming_decode.py:582] Cuts processed until now is 6700.
173
+ 2025-11-17 17:38:46,244 INFO [streaming_decode.py:582] Cuts processed until now is 6800.
174
+ 2025-11-17 17:38:47,688 INFO [streaming_decode.py:582] Cuts processed until now is 6900.
175
+ 2025-11-17 17:38:49,033 INFO [streaming_decode.py:582] Cuts processed until now is 7000.
176
+ 2025-11-17 17:38:50,527 INFO [streaming_decode.py:582] Cuts processed until now is 7100.
177
+ 2025-11-17 17:38:51,965 INFO [streaming_decode.py:582] Cuts processed until now is 7200.
178
+ 2025-11-17 17:38:53,372 INFO [streaming_decode.py:582] Cuts processed until now is 7300.
179
+ 2025-11-17 17:38:54,690 INFO [streaming_decode.py:582] Cuts processed until now is 7400.
180
+ 2025-11-17 17:38:56,009 INFO [streaming_decode.py:582] Cuts processed until now is 7500.
181
+ 2025-11-17 17:38:57,289 INFO [streaming_decode.py:582] Cuts processed until now is 7600.
182
+ 2025-11-17 17:38:58,453 INFO [streaming_decode.py:582] Cuts processed until now is 7700.
183
+ 2025-11-17 17:38:59,828 INFO [streaming_decode.py:582] Cuts processed until now is 7800.
184
+ 2025-11-17 17:39:01,214 INFO [streaming_decode.py:582] Cuts processed until now is 7900.
185
+ 2025-11-17 17:39:02,689 INFO [streaming_decode.py:582] Cuts processed until now is 8000.
186
+ 2025-11-17 17:39:04,355 INFO [streaming_decode.py:582] Cuts processed until now is 8100.
187
+ 2025-11-17 17:39:05,642 INFO [streaming_decode.py:582] Cuts processed until now is 8200.
188
+ 2025-11-17 17:39:07,031 INFO [streaming_decode.py:582] Cuts processed until now is 8300.
189
+ 2025-11-17 17:39:08,222 INFO [streaming_decode.py:582] Cuts processed until now is 8400.
190
+ 2025-11-17 17:39:09,675 INFO [streaming_decode.py:582] Cuts processed until now is 8500.
191
+ 2025-11-17 17:39:10,922 INFO [streaming_decode.py:582] Cuts processed until now is 8600.
192
+ 2025-11-17 17:39:12,088 INFO [streaming_decode.py:582] Cuts processed until now is 8700.
193
+ 2025-11-17 17:39:13,272 INFO [streaming_decode.py:582] Cuts processed until now is 8800.
194
+ 2025-11-17 17:39:14,613 INFO [streaming_decode.py:582] Cuts processed until now is 8900.
195
+ 2025-11-17 17:39:15,988 INFO [streaming_decode.py:582] Cuts processed until now is 9000.
196
+ 2025-11-17 17:39:17,266 INFO [streaming_decode.py:582] Cuts processed until now is 9100.
197
+ 2025-11-17 17:39:18,574 INFO [streaming_decode.py:582] Cuts processed until now is 9200.
198
+ 2025-11-17 17:39:19,706 INFO [streaming_decode.py:582] Cuts processed until now is 9300.
199
+ 2025-11-17 17:39:21,085 INFO [streaming_decode.py:582] Cuts processed until now is 9400.
200
+ 2025-11-17 17:39:22,276 INFO [streaming_decode.py:582] Cuts processed until now is 9500.
201
+ 2025-11-17 17:39:23,742 INFO [streaming_decode.py:582] Cuts processed until now is 9600.
202
+ 2025-11-17 17:39:24,985 INFO [streaming_decode.py:582] Cuts processed until now is 9700.
203
+ 2025-11-17 17:39:26,319 INFO [streaming_decode.py:582] Cuts processed until now is 9800.
204
+ 2025-11-17 17:39:27,671 INFO [streaming_decode.py:582] Cuts processed until now is 9900.
205
+ 2025-11-17 17:39:28,952 INFO [streaming_decode.py:582] Cuts processed until now is 10000.
206
+ 2025-11-17 17:39:30,204 INFO [streaming_decode.py:582] Cuts processed until now is 10100.
207
+ 2025-11-17 17:39:31,542 INFO [streaming_decode.py:582] Cuts processed until now is 10200.
208
+ 2025-11-17 17:39:32,876 INFO [streaming_decode.py:582] Cuts processed until now is 10300.
209
+ 2025-11-17 17:39:34,152 INFO [streaming_decode.py:582] Cuts processed until now is 10400.
210
+ 2025-11-17 17:39:35,826 INFO [streaming_decode.py:582] Cuts processed until now is 10500.
211
+ 2025-11-17 17:39:36,959 INFO [streaming_decode.py:582] Cuts processed until now is 10600.
212
+ 2025-11-17 17:39:38,027 INFO [streaming_decode.py:582] Cuts processed until now is 10700.
213
+ 2025-11-17 17:39:39,410 INFO [streaming_decode.py:582] Cuts processed until now is 10800.
214
+ 2025-11-17 17:39:40,850 INFO [streaming_decode.py:582] Cuts processed until now is 10900.
215
+ 2025-11-17 17:39:42,571 INFO [streaming_decode.py:582] Cuts processed until now is 11000.
216
+ 2025-11-17 17:39:44,007 INFO [streaming_decode.py:582] Cuts processed until now is 11100.
217
+ 2025-11-17 17:39:45,436 INFO [streaming_decode.py:582] Cuts processed until now is 11200.
218
+ 2025-11-17 17:39:47,010 INFO [streaming_decode.py:582] Cuts processed until now is 11300.
219
+ 2025-11-17 17:39:48,521 INFO [streaming_decode.py:582] Cuts processed until now is 11400.
220
+ 2025-11-17 17:39:50,182 INFO [streaming_decode.py:582] Cuts processed until now is 11500.
221
+ 2025-11-17 17:39:51,374 INFO [streaming_decode.py:582] Cuts processed until now is 11600.
222
+ 2025-11-17 17:39:52,962 INFO [streaming_decode.py:582] Cuts processed until now is 11700.
223
+ 2025-11-17 17:39:54,096 INFO [streaming_decode.py:582] Cuts processed until now is 11800.
224
+ 2025-11-17 17:39:55,127 INFO [streaming_decode.py:582] Cuts processed until now is 11900.
225
+ 2025-11-17 17:39:57,036 INFO [streaming_decode.py:582] Cuts processed until now is 12000.
226
+ 2025-11-17 17:39:58,503 INFO [streaming_decode.py:582] Cuts processed until now is 12100.
227
+ 2025-11-17 17:40:00,115 INFO [streaming_decode.py:582] Cuts processed until now is 12200.
228
+ 2025-11-17 17:40:01,289 INFO [streaming_decode.py:582] Cuts processed until now is 12300.
229
+ 2025-11-17 17:40:02,645 INFO [streaming_decode.py:582] Cuts processed until now is 12400.
230
+ 2025-11-17 17:40:03,923 INFO [streaming_decode.py:582] Cuts processed until now is 12500.
231
+ 2025-11-17 17:40:05,412 INFO [streaming_decode.py:582] Cuts processed until now is 12600.
232
+ 2025-11-17 17:40:06,638 INFO [streaming_decode.py:582] Cuts processed until now is 12700.
233
+ 2025-11-17 17:40:07,937 INFO [streaming_decode.py:582] Cuts processed until now is 12800.
234
+ 2025-11-17 17:40:09,320 INFO [streaming_decode.py:582] Cuts processed until now is 12900.
235
+ 2025-11-17 17:40:11,096 INFO [streaming_decode.py:582] Cuts processed until now is 13000.
236
+ 2025-11-17 17:40:12,554 INFO [streaming_decode.py:582] Cuts processed until now is 13100.
237
+ 2025-11-17 17:40:13,922 INFO [streaming_decode.py:582] Cuts processed until now is 13200.
238
+ 2025-11-17 17:40:15,192 INFO [streaming_decode.py:582] Cuts processed until now is 13300.
239
+ 2025-11-17 17:40:16,561 INFO [streaming_decode.py:582] Cuts processed until now is 13400.
240
+ 2025-11-17 17:40:17,904 INFO [streaming_decode.py:582] Cuts processed until now is 13500.
241
+ 2025-11-17 17:40:19,107 INFO [streaming_decode.py:582] Cuts processed until now is 13600.
242
+ 2025-11-17 17:40:20,792 INFO [streaming_decode.py:582] Cuts processed until now is 13700.
243
+ 2025-11-17 17:40:21,997 INFO [streaming_decode.py:582] Cuts processed until now is 13800.
244
+ 2025-11-17 17:40:23,234 INFO [streaming_decode.py:582] Cuts processed until now is 13900.
245
+ 2025-11-17 17:40:24,656 INFO [streaming_decode.py:582] Cuts processed until now is 14000.
246
+ 2025-11-17 17:40:26,159 INFO [streaming_decode.py:582] Cuts processed until now is 14100.
247
+ 2025-11-17 17:40:27,955 INFO [streaming_decode.py:582] Cuts processed until now is 14200.
248
+ 2025-11-17 17:40:29,388 INFO [streaming_decode.py:582] Cuts processed until now is 14300.
249
+ 2025-11-17 17:40:30,606 INFO [streaming_decode.py:582] Cuts processed until now is 14400.
250
+ 2025-11-17 17:40:31,863 INFO [streaming_decode.py:582] Cuts processed until now is 14500.
251
+ 2025-11-17 17:40:33,354 INFO [streaming_decode.py:582] Cuts processed until now is 14600.
252
+ 2025-11-17 17:40:34,878 INFO [streaming_decode.py:582] Cuts processed until now is 14700.
253
+ 2025-11-17 17:40:36,166 INFO [streaming_decode.py:582] Cuts processed until now is 14800.
254
+ 2025-11-17 17:40:37,476 INFO [streaming_decode.py:582] Cuts processed until now is 14900.
255
+ 2025-11-17 17:40:38,733 INFO [streaming_decode.py:582] Cuts processed until now is 15000.
256
+ 2025-11-17 17:40:40,107 INFO [streaming_decode.py:582] Cuts processed until now is 15100.
257
+ 2025-11-17 17:40:41,326 INFO [streaming_decode.py:582] Cuts processed until now is 15200.
258
+ 2025-11-17 17:40:42,825 INFO [streaming_decode.py:582] Cuts processed until now is 15300.
259
+ 2025-11-17 17:40:44,214 INFO [streaming_decode.py:582] Cuts processed until now is 15400.
260
+ 2025-11-17 17:40:45,705 INFO [streaming_decode.py:582] Cuts processed until now is 15500.
261
+ 2025-11-17 17:40:47,256 INFO [streaming_decode.py:582] Cuts processed until now is 15600.
262
+ 2025-11-17 17:40:48,503 INFO [streaming_decode.py:582] Cuts processed until now is 15700.
263
+ 2025-11-17 17:40:49,914 INFO [streaming_decode.py:582] Cuts processed until now is 15800.
264
+ 2025-11-17 17:40:55,096 INFO [streaming_decode.py:618] The transcripts are stored in tmp/exp-causal-80-epoch/streaming/greedy_search/recogs-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt
265
+ 2025-11-17 17:40:55,394 INFO [utils.py:670] [test-commonvoice-greedy_search] %WER 2.85% [22183 / 778666, 3660 ins, 9528 del, 8995 sub ]
266
+ 2025-11-17 17:40:56,240 INFO [streaming_decode.py:627] Wrote detailed error stats to tmp/exp-causal-80-epoch/streaming/greedy_search/errs-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt
267
+ 2025-11-17 17:40:56,241 INFO [streaming_decode.py:641]
268
+ For test-commonvoice, WER of different settings are:
269
+ greedy_search 2.85 best for test-commonvoice
270
+
271
+ 2025-11-17 17:40:56,249 INFO [streaming_decode.py:582] Cuts processed until now is 0.
272
+ 2025-11-17 17:40:57,004 INFO [streaming_decode.py:582] Cuts processed until now is 100.
273
+ 2025-11-17 17:40:57,637 INFO [streaming_decode.py:582] Cuts processed until now is 200.
274
+ 2025-11-17 17:40:58,548 INFO [streaming_decode.py:582] Cuts processed until now is 300.
275
+ 2025-11-17 17:40:59,294 INFO [streaming_decode.py:582] Cuts processed until now is 400.
276
+ 2025-11-17 17:41:00,158 INFO [streaming_decode.py:582] Cuts processed until now is 500.
277
+ 2025-11-17 17:41:01,145 INFO [streaming_decode.py:582] Cuts processed until now is 600.
278
+ 2025-11-17 17:41:02,058 INFO [streaming_decode.py:582] Cuts processed until now is 700.
279
+ 2025-11-17 17:41:02,704 INFO [streaming_decode.py:582] Cuts processed until now is 800.
280
+ 2025-11-17 17:41:03,343 INFO [streaming_decode.py:582] Cuts processed until now is 900.
281
+ 2025-11-17 17:41:09,300 INFO [streaming_decode.py:618] The transcripts are stored in tmp/exp-causal-80-epoch/streaming/greedy_search/recogs-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt
282
+ 2025-11-17 17:41:09,316 INFO [utils.py:670] [test-slr72-greedy_search] %WER 1.56% [635 / 40600, 94 ins, 360 del, 181 sub ]
283
+ 2025-11-17 17:41:09,357 INFO [streaming_decode.py:627] Wrote detailed error stats to tmp/exp-causal-80-epoch/streaming/greedy_search/errs-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt
284
+ 2025-11-17 17:41:09,357 INFO [streaming_decode.py:641]
285
+ For test-slr72, WER of different settings are:
286
+ greedy_search 1.56 best for test-slr72
287
+
288
+ 2025-11-17 17:41:09,357 INFO [streaming_decode.py:794] Done!
streaming/greedy_search/log-decode-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model-2025-11-17-16-48-52 ADDED
@@ -0,0 +1,114 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2025-11-17 16:48:52,315 INFO [streaming_decode.py:677] Decoding started
2
+ 2025-11-17 16:48:52,315 INFO [streaming_decode.py:683] Device: cuda:0
3
+ 2025-11-17 16:48:52,316 INFO [lexicon.py:168] Loading pre-compiled data/lang_phone/Linv.pt
4
+ 2025-11-17 16:48:52,316 INFO [streaming_decode.py:691] {
5
+ "attention_decoder_attention_dim": 512,
6
+ "attention_decoder_dim": 512,
7
+ "attention_decoder_feedforward_dim": 2048,
8
+ "attention_decoder_num_heads": 8,
9
+ "attention_decoder_num_layers": 6,
10
+ "avg": 5,
11
+ "batch_idx_train": 0,
12
+ "beam": 4,
13
+ "best_train_epoch": -1,
14
+ "best_train_loss": Infinity,
15
+ "best_valid_epoch": -1,
16
+ "best_valid_loss": Infinity,
17
+ "blank_id": 0,
18
+ "bucketing_sampler": true,
19
+ "causal": true,
20
+ "chunk_size": "32",
21
+ "cnn_module_kernel": "31,31,15,15,15,31",
22
+ "concatenate_cuts": false,
23
+ "context_size": 2,
24
+ "decoder_dim": 512,
25
+ "decoding_method": "greedy_search",
26
+ "downsampling_factor": "1,2,4,8,4,2",
27
+ "drop_last": true,
28
+ "duration_factor": 1.0,
29
+ "enable_musan": true,
30
+ "enable_spec_aug": true,
31
+ "encoder_dim": "192,256,256,256,256,256",
32
+ "encoder_unmasked_dim": "192,192,192,192,192,192",
33
+ "env_info": {
34
+ "IP address": "127.0.1.1",
35
+ "hostname": "Bookbot-GPU1",
36
+ "icefall-git-branch": "master",
37
+ "icefall-git-date": "Tue Jan 7 16:20:44 2025",
38
+ "icefall-git-sha1": "77cd018f-dirty",
39
+ "icefall-path": "/mnt/Projects/Projects/ASR/icefall",
40
+ "k2-build-type": "Release",
41
+ "k2-git-date": "Thu Jul 25 03:46:03 2024",
42
+ "k2-git-sha1": "5735fa707f6091856d13ccd230aced6e9e64f815",
43
+ "k2-path": "/home/bookbot/miniconda3/envs/icefall/lib/python3.12/site-packages/k2/__init__.py",
44
+ "k2-version": "1.24.4",
45
+ "k2-with-cuda": true,
46
+ "lhotse-path": "/home/bookbot/miniconda3/envs/icefall/lib/python3.12/site-packages/lhotse/__init__.py",
47
+ "lhotse-version": "1.31.1.dev+git.4c43958.clean",
48
+ "python-version": "3.12",
49
+ "torch-cuda-available": true,
50
+ "torch-cuda-version": "12.4",
51
+ "torch-version": "2.4.0+cu124"
52
+ },
53
+ "epoch": 80,
54
+ "exp_dir": "tmp/exp-causal-80-epoch",
55
+ "feature_dim": 80,
56
+ "feedforward_dim": "512,768,768,768,768,768",
57
+ "gap": 1.0,
58
+ "ignore_id": -1,
59
+ "input_strategy": "PrecomputedFeatures",
60
+ "iter": 0,
61
+ "joiner_dim": 512,
62
+ "label_smoothing": 0.1,
63
+ "lang_dir": "data/lang_phone",
64
+ "left_context_frames": "128",
65
+ "log_interval": 50,
66
+ "manifest_dir": "data/fbank",
67
+ "max_contexts": 4,
68
+ "max_duration": 200.0,
69
+ "max_states": 32,
70
+ "num_active_paths": 4,
71
+ "num_buckets": 30,
72
+ "num_decode_streams": 1000,
73
+ "num_encoder_layers": "2,2,2,2,2,2",
74
+ "num_heads": "4,4,4,8,4,4",
75
+ "num_workers": 2,
76
+ "on_the_fly_feats": false,
77
+ "pos_dim": 48,
78
+ "pos_head_dim": "4",
79
+ "query_head_dim": "32",
80
+ "res_dir": "tmp/exp-causal-80-epoch/streaming/greedy_search",
81
+ "reset_interval": 200,
82
+ "return_cuts": true,
83
+ "shuffle": true,
84
+ "spec_aug_time_warp_factor": 80,
85
+ "subsampling_factor": 4,
86
+ "suffix": "epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model",
87
+ "unk_id": 1,
88
+ "use_attention_decoder": false,
89
+ "use_averaged_model": true,
90
+ "use_cr_ctc": false,
91
+ "use_ctc": false,
92
+ "use_transducer": true,
93
+ "valid_interval": 3000,
94
+ "value_head_dim": "12",
95
+ "vocab_size": 36,
96
+ "warm_step": 2000
97
+ }
98
+ 2025-11-17 16:48:52,317 INFO [streaming_decode.py:693] About to create model
99
+ 2025-11-17 16:48:52,516 INFO [streaming_decode.py:748] Calculating the averaged model over epoch range from 75 (excluded) to 80
100
+ 2025-11-17 16:48:56,055 INFO [streaming_decode.py:769] Number of model parameters: 22795007
101
+ 2025-11-17 16:48:56,056 INFO [multidataset.py:83] About to get Spanish Common Voice test cuts
102
+ 2025-11-17 16:48:56,056 INFO [multidataset.py:85] Loading Spanish Common Voice in lazy mode
103
+ 2025-11-17 16:48:56,056 INFO [multidataset.py:94] About to get SLR72 dataset test cuts
104
+ 2025-11-17 16:48:56,056 INFO [multidataset.py:96] Loading SLR72 dataset in lazy mode
105
+ 2025-11-17 16:48:56,168 INFO [streaming_decode.py:582] Cuts processed until now is 0.
106
+ 2025-11-17 16:48:58,537 INFO [streaming_decode.py:582] Cuts processed until now is 100.
107
+ 2025-11-17 16:49:01,695 INFO [streaming_decode.py:582] Cuts processed until now is 200.
108
+ 2025-11-17 16:49:04,078 INFO [streaming_decode.py:582] Cuts processed until now is 300.
109
+ 2025-11-17 16:49:05,726 INFO [streaming_decode.py:582] Cuts processed until now is 400.
110
+ 2025-11-17 16:49:08,259 INFO [streaming_decode.py:582] Cuts processed until now is 500.
111
+ 2025-11-17 16:49:10,437 INFO [streaming_decode.py:582] Cuts processed until now is 600.
112
+ 2025-11-17 16:49:12,978 INFO [streaming_decode.py:582] Cuts processed until now is 700.
113
+ 2025-11-17 16:49:15,156 INFO [streaming_decode.py:582] Cuts processed until now is 800.
114
+ 2025-11-17 16:49:18,085 INFO [streaming_decode.py:582] Cuts processed until now is 900.
streaming/greedy_search/log-decode-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model-2025-11-17-16-55-04 ADDED
@@ -0,0 +1,288 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2025-11-17 16:55:04,359 INFO [streaming_decode.py:677] Decoding started
2
+ 2025-11-17 16:55:04,359 INFO [streaming_decode.py:683] Device: cuda:0
3
+ 2025-11-17 16:55:04,359 INFO [lexicon.py:168] Loading pre-compiled data/lang_phone/Linv.pt
4
+ 2025-11-17 16:55:04,360 INFO [streaming_decode.py:691] {
5
+ "attention_decoder_attention_dim": 512,
6
+ "attention_decoder_dim": 512,
7
+ "attention_decoder_feedforward_dim": 2048,
8
+ "attention_decoder_num_heads": 8,
9
+ "attention_decoder_num_layers": 6,
10
+ "avg": 5,
11
+ "batch_idx_train": 0,
12
+ "beam": 4,
13
+ "best_train_epoch": -1,
14
+ "best_train_loss": Infinity,
15
+ "best_valid_epoch": -1,
16
+ "best_valid_loss": Infinity,
17
+ "blank_id": 0,
18
+ "bucketing_sampler": true,
19
+ "causal": true,
20
+ "chunk_size": "32",
21
+ "cnn_module_kernel": "31,31,15,15,15,31",
22
+ "concatenate_cuts": false,
23
+ "context_size": 2,
24
+ "decoder_dim": 512,
25
+ "decoding_method": "greedy_search",
26
+ "downsampling_factor": "1,2,4,8,4,2",
27
+ "drop_last": true,
28
+ "duration_factor": 1.0,
29
+ "enable_musan": true,
30
+ "enable_spec_aug": true,
31
+ "encoder_dim": "192,256,256,256,256,256",
32
+ "encoder_unmasked_dim": "192,192,192,192,192,192",
33
+ "env_info": {
34
+ "IP address": "127.0.1.1",
35
+ "hostname": "Bookbot-GPU1",
36
+ "icefall-git-branch": "master",
37
+ "icefall-git-date": "Tue Jan 7 16:20:44 2025",
38
+ "icefall-git-sha1": "77cd018f-dirty",
39
+ "icefall-path": "/mnt/Projects/Projects/ASR/icefall",
40
+ "k2-build-type": "Release",
41
+ "k2-git-date": "Thu Jul 25 03:46:03 2024",
42
+ "k2-git-sha1": "5735fa707f6091856d13ccd230aced6e9e64f815",
43
+ "k2-path": "/home/bookbot/miniconda3/envs/icefall/lib/python3.12/site-packages/k2/__init__.py",
44
+ "k2-version": "1.24.4",
45
+ "k2-with-cuda": true,
46
+ "lhotse-path": "/home/bookbot/miniconda3/envs/icefall/lib/python3.12/site-packages/lhotse/__init__.py",
47
+ "lhotse-version": "1.31.1.dev+git.4c43958.clean",
48
+ "python-version": "3.12",
49
+ "torch-cuda-available": true,
50
+ "torch-cuda-version": "12.4",
51
+ "torch-version": "2.4.0+cu124"
52
+ },
53
+ "epoch": 80,
54
+ "exp_dir": "tmp/exp-causal-80-epoch",
55
+ "feature_dim": 80,
56
+ "feedforward_dim": "512,768,768,768,768,768",
57
+ "gap": 1.0,
58
+ "ignore_id": -1,
59
+ "input_strategy": "PrecomputedFeatures",
60
+ "iter": 0,
61
+ "joiner_dim": 512,
62
+ "label_smoothing": 0.1,
63
+ "lang_dir": "data/lang_phone",
64
+ "left_context_frames": "128",
65
+ "log_interval": 50,
66
+ "manifest_dir": "data/fbank",
67
+ "max_contexts": 4,
68
+ "max_duration": 200.0,
69
+ "max_states": 32,
70
+ "num_active_paths": 4,
71
+ "num_buckets": 30,
72
+ "num_decode_streams": 1000,
73
+ "num_encoder_layers": "2,2,2,2,2,2",
74
+ "num_heads": "4,4,4,8,4,4",
75
+ "num_workers": 2,
76
+ "on_the_fly_feats": false,
77
+ "pos_dim": 48,
78
+ "pos_head_dim": "4",
79
+ "query_head_dim": "32",
80
+ "res_dir": "tmp/exp-causal-80-epoch/streaming/greedy_search",
81
+ "reset_interval": 200,
82
+ "return_cuts": true,
83
+ "shuffle": true,
84
+ "spec_aug_time_warp_factor": 80,
85
+ "subsampling_factor": 4,
86
+ "suffix": "epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model",
87
+ "unk_id": 1,
88
+ "use_attention_decoder": false,
89
+ "use_averaged_model": true,
90
+ "use_cr_ctc": false,
91
+ "use_ctc": false,
92
+ "use_transducer": true,
93
+ "valid_interval": 3000,
94
+ "value_head_dim": "12",
95
+ "vocab_size": 36,
96
+ "warm_step": 2000
97
+ }
98
+ 2025-11-17 16:55:04,361 INFO [streaming_decode.py:693] About to create model
99
+ 2025-11-17 16:55:04,561 INFO [streaming_decode.py:748] Calculating the averaged model over epoch range from 75 (excluded) to 80
100
+ 2025-11-17 16:55:05,336 INFO [streaming_decode.py:769] Number of model parameters: 22795007
101
+ 2025-11-17 16:55:05,336 INFO [multidataset.py:83] About to get Spanish Common Voice test cuts
102
+ 2025-11-17 16:55:05,336 INFO [multidataset.py:85] Loading Spanish Common Voice in lazy mode
103
+ 2025-11-17 16:55:05,337 INFO [multidataset.py:94] About to get SLR72 dataset test cuts
104
+ 2025-11-17 16:55:05,337 INFO [multidataset.py:96] Loading SLR72 dataset in lazy mode
105
+ 2025-11-17 16:55:05,522 INFO [streaming_decode.py:582] Cuts processed until now is 0.
106
+ 2025-11-17 16:55:06,436 INFO [streaming_decode.py:582] Cuts processed until now is 100.
107
+ 2025-11-17 16:55:07,184 INFO [streaming_decode.py:582] Cuts processed until now is 200.
108
+ 2025-11-17 16:55:08,234 INFO [streaming_decode.py:582] Cuts processed until now is 300.
109
+ 2025-11-17 16:55:08,835 INFO [streaming_decode.py:582] Cuts processed until now is 400.
110
+ 2025-11-17 16:55:09,589 INFO [streaming_decode.py:582] Cuts processed until now is 500.
111
+ 2025-11-17 16:55:10,745 INFO [streaming_decode.py:582] Cuts processed until now is 600.
112
+ 2025-11-17 16:55:11,429 INFO [streaming_decode.py:582] Cuts processed until now is 700.
113
+ 2025-11-17 16:55:12,138 INFO [streaming_decode.py:582] Cuts processed until now is 800.
114
+ 2025-11-17 16:55:13,048 INFO [streaming_decode.py:582] Cuts processed until now is 900.
115
+ 2025-11-17 16:55:15,600 INFO [streaming_decode.py:582] Cuts processed until now is 1000.
116
+ 2025-11-17 16:55:17,648 INFO [streaming_decode.py:582] Cuts processed until now is 1100.
117
+ 2025-11-17 16:55:18,796 INFO [streaming_decode.py:582] Cuts processed until now is 1200.
118
+ 2025-11-17 16:55:20,019 INFO [streaming_decode.py:582] Cuts processed until now is 1300.
119
+ 2025-11-17 16:55:21,506 INFO [streaming_decode.py:582] Cuts processed until now is 1400.
120
+ 2025-11-17 16:55:22,264 INFO [streaming_decode.py:582] Cuts processed until now is 1500.
121
+ 2025-11-17 16:55:23,461 INFO [streaming_decode.py:582] Cuts processed until now is 1600.
122
+ 2025-11-17 16:55:24,748 INFO [streaming_decode.py:582] Cuts processed until now is 1700.
123
+ 2025-11-17 16:55:26,087 INFO [streaming_decode.py:582] Cuts processed until now is 1800.
124
+ 2025-11-17 16:55:27,321 INFO [streaming_decode.py:582] Cuts processed until now is 1900.
125
+ 2025-11-17 16:55:28,785 INFO [streaming_decode.py:582] Cuts processed until now is 2000.
126
+ 2025-11-17 16:55:29,875 INFO [streaming_decode.py:582] Cuts processed until now is 2100.
127
+ 2025-11-17 16:55:31,134 INFO [streaming_decode.py:582] Cuts processed until now is 2200.
128
+ 2025-11-17 16:55:32,215 INFO [streaming_decode.py:582] Cuts processed until now is 2300.
129
+ 2025-11-17 16:55:33,472 INFO [streaming_decode.py:582] Cuts processed until now is 2400.
130
+ 2025-11-17 16:55:34,510 INFO [streaming_decode.py:582] Cuts processed until now is 2500.
131
+ 2025-11-17 16:55:35,610 INFO [streaming_decode.py:582] Cuts processed until now is 2600.
132
+ 2025-11-17 16:55:36,640 INFO [streaming_decode.py:582] Cuts processed until now is 2700.
133
+ 2025-11-17 16:55:37,814 INFO [streaming_decode.py:582] Cuts processed until now is 2800.
134
+ 2025-11-17 16:55:39,084 INFO [streaming_decode.py:582] Cuts processed until now is 2900.
135
+ 2025-11-17 16:55:40,137 INFO [streaming_decode.py:582] Cuts processed until now is 3000.
136
+ 2025-11-17 16:55:41,829 INFO [streaming_decode.py:582] Cuts processed until now is 3100.
137
+ 2025-11-17 16:55:42,884 INFO [streaming_decode.py:582] Cuts processed until now is 3200.
138
+ 2025-11-17 16:55:44,028 INFO [streaming_decode.py:582] Cuts processed until now is 3300.
139
+ 2025-11-17 16:55:45,475 INFO [streaming_decode.py:582] Cuts processed until now is 3400.
140
+ 2025-11-17 16:55:46,251 INFO [streaming_decode.py:582] Cuts processed until now is 3500.
141
+ 2025-11-17 16:55:47,293 INFO [streaming_decode.py:582] Cuts processed until now is 3600.
142
+ 2025-11-17 16:55:48,546 INFO [streaming_decode.py:582] Cuts processed until now is 3700.
143
+ 2025-11-17 16:55:49,670 INFO [streaming_decode.py:582] Cuts processed until now is 3800.
144
+ 2025-11-17 16:55:50,572 INFO [streaming_decode.py:582] Cuts processed until now is 3900.
145
+ 2025-11-17 16:55:51,493 INFO [streaming_decode.py:582] Cuts processed until now is 4000.
146
+ 2025-11-17 16:55:52,750 INFO [streaming_decode.py:582] Cuts processed until now is 4100.
147
+ 2025-11-17 16:55:53,825 INFO [streaming_decode.py:582] Cuts processed until now is 4200.
148
+ 2025-11-17 16:55:54,975 INFO [streaming_decode.py:582] Cuts processed until now is 4300.
149
+ 2025-11-17 16:55:55,902 INFO [streaming_decode.py:582] Cuts processed until now is 4400.
150
+ 2025-11-17 16:55:57,112 INFO [streaming_decode.py:582] Cuts processed until now is 4500.
151
+ 2025-11-17 16:55:58,152 INFO [streaming_decode.py:582] Cuts processed until now is 4600.
152
+ 2025-11-17 16:55:59,414 INFO [streaming_decode.py:582] Cuts processed until now is 4700.
153
+ 2025-11-17 16:56:00,860 INFO [streaming_decode.py:582] Cuts processed until now is 4800.
154
+ 2025-11-17 16:56:02,124 INFO [streaming_decode.py:582] Cuts processed until now is 4900.
155
+ 2025-11-17 16:56:03,246 INFO [streaming_decode.py:582] Cuts processed until now is 5000.
156
+ 2025-11-17 16:56:04,269 INFO [streaming_decode.py:582] Cuts processed until now is 5100.
157
+ 2025-11-17 16:56:05,459 INFO [streaming_decode.py:582] Cuts processed until now is 5200.
158
+ 2025-11-17 16:56:06,395 INFO [streaming_decode.py:582] Cuts processed until now is 5300.
159
+ 2025-11-17 16:56:07,632 INFO [streaming_decode.py:582] Cuts processed until now is 5400.
160
+ 2025-11-17 16:56:08,746 INFO [streaming_decode.py:582] Cuts processed until now is 5500.
161
+ 2025-11-17 16:56:09,785 INFO [streaming_decode.py:582] Cuts processed until now is 5600.
162
+ 2025-11-17 16:56:10,969 INFO [streaming_decode.py:582] Cuts processed until now is 5700.
163
+ 2025-11-17 16:56:12,109 INFO [streaming_decode.py:582] Cuts processed until now is 5800.
164
+ 2025-11-17 16:56:13,544 INFO [streaming_decode.py:582] Cuts processed until now is 5900.
165
+ 2025-11-17 16:56:14,559 INFO [streaming_decode.py:582] Cuts processed until now is 6000.
166
+ 2025-11-17 16:56:15,827 INFO [streaming_decode.py:582] Cuts processed until now is 6100.
167
+ 2025-11-17 16:56:16,828 INFO [streaming_decode.py:582] Cuts processed until now is 6200.
168
+ 2025-11-17 16:56:17,630 INFO [streaming_decode.py:582] Cuts processed until now is 6300.
169
+ 2025-11-17 16:56:18,609 INFO [streaming_decode.py:582] Cuts processed until now is 6400.
170
+ 2025-11-17 16:56:19,833 INFO [streaming_decode.py:582] Cuts processed until now is 6500.
171
+ 2025-11-17 16:56:20,944 INFO [streaming_decode.py:582] Cuts processed until now is 6600.
172
+ 2025-11-17 16:56:22,267 INFO [streaming_decode.py:582] Cuts processed until now is 6700.
173
+ 2025-11-17 16:56:23,186 INFO [streaming_decode.py:582] Cuts processed until now is 6800.
174
+ 2025-11-17 16:56:24,401 INFO [streaming_decode.py:582] Cuts processed until now is 6900.
175
+ 2025-11-17 16:56:25,468 INFO [streaming_decode.py:582] Cuts processed until now is 7000.
176
+ 2025-11-17 16:56:26,425 INFO [streaming_decode.py:582] Cuts processed until now is 7100.
177
+ 2025-11-17 16:56:27,997 INFO [streaming_decode.py:582] Cuts processed until now is 7200.
178
+ 2025-11-17 16:56:29,116 INFO [streaming_decode.py:582] Cuts processed until now is 7300.
179
+ 2025-11-17 16:56:30,196 INFO [streaming_decode.py:582] Cuts processed until now is 7400.
180
+ 2025-11-17 16:56:31,473 INFO [streaming_decode.py:582] Cuts processed until now is 7500.
181
+ 2025-11-17 16:56:32,817 INFO [streaming_decode.py:582] Cuts processed until now is 7600.
182
+ 2025-11-17 16:56:33,700 INFO [streaming_decode.py:582] Cuts processed until now is 7700.
183
+ 2025-11-17 16:56:34,925 INFO [streaming_decode.py:582] Cuts processed until now is 7800.
184
+ 2025-11-17 16:56:36,247 INFO [streaming_decode.py:582] Cuts processed until now is 7900.
185
+ 2025-11-17 16:56:37,592 INFO [streaming_decode.py:582] Cuts processed until now is 8000.
186
+ 2025-11-17 16:56:38,617 INFO [streaming_decode.py:582] Cuts processed until now is 8100.
187
+ 2025-11-17 16:56:39,811 INFO [streaming_decode.py:582] Cuts processed until now is 8200.
188
+ 2025-11-17 16:56:40,945 INFO [streaming_decode.py:582] Cuts processed until now is 8300.
189
+ 2025-11-17 16:56:41,915 INFO [streaming_decode.py:582] Cuts processed until now is 8400.
190
+ 2025-11-17 16:56:42,885 INFO [streaming_decode.py:582] Cuts processed until now is 8500.
191
+ 2025-11-17 16:56:43,862 INFO [streaming_decode.py:582] Cuts processed until now is 8600.
192
+ 2025-11-17 16:56:45,145 INFO [streaming_decode.py:582] Cuts processed until now is 8700.
193
+ 2025-11-17 16:56:46,668 INFO [streaming_decode.py:582] Cuts processed until now is 8800.
194
+ 2025-11-17 16:56:47,773 INFO [streaming_decode.py:582] Cuts processed until now is 8900.
195
+ 2025-11-17 16:56:49,097 INFO [streaming_decode.py:582] Cuts processed until now is 9000.
196
+ 2025-11-17 16:56:50,256 INFO [streaming_decode.py:582] Cuts processed until now is 9100.
197
+ 2025-11-17 16:56:51,566 INFO [streaming_decode.py:582] Cuts processed until now is 9200.
198
+ 2025-11-17 16:56:52,581 INFO [streaming_decode.py:582] Cuts processed until now is 9300.
199
+ 2025-11-17 16:56:53,804 INFO [streaming_decode.py:582] Cuts processed until now is 9400.
200
+ 2025-11-17 16:56:54,805 INFO [streaming_decode.py:582] Cuts processed until now is 9500.
201
+ 2025-11-17 16:56:55,807 INFO [streaming_decode.py:582] Cuts processed until now is 9600.
202
+ 2025-11-17 16:56:57,140 INFO [streaming_decode.py:582] Cuts processed until now is 9700.
203
+ 2025-11-17 16:56:58,552 INFO [streaming_decode.py:582] Cuts processed until now is 9800.
204
+ 2025-11-17 16:56:59,879 INFO [streaming_decode.py:582] Cuts processed until now is 9900.
205
+ 2025-11-17 16:57:00,856 INFO [streaming_decode.py:582] Cuts processed until now is 10000.
206
+ 2025-11-17 16:57:02,089 INFO [streaming_decode.py:582] Cuts processed until now is 10100.
207
+ 2025-11-17 16:57:03,338 INFO [streaming_decode.py:582] Cuts processed until now is 10200.
208
+ 2025-11-17 16:57:04,990 INFO [streaming_decode.py:582] Cuts processed until now is 10300.
209
+ 2025-11-17 16:57:06,143 INFO [streaming_decode.py:582] Cuts processed until now is 10400.
210
+ 2025-11-17 16:57:07,181 INFO [streaming_decode.py:582] Cuts processed until now is 10500.
211
+ 2025-11-17 16:57:08,488 INFO [streaming_decode.py:582] Cuts processed until now is 10600.
212
+ 2025-11-17 16:57:09,659 INFO [streaming_decode.py:582] Cuts processed until now is 10700.
213
+ 2025-11-17 16:57:10,927 INFO [streaming_decode.py:582] Cuts processed until now is 10800.
214
+ 2025-11-17 16:57:11,921 INFO [streaming_decode.py:582] Cuts processed until now is 10900.
215
+ 2025-11-17 16:57:13,038 INFO [streaming_decode.py:582] Cuts processed until now is 11000.
216
+ 2025-11-17 16:57:13,939 INFO [streaming_decode.py:582] Cuts processed until now is 11100.
217
+ 2025-11-17 16:57:15,151 INFO [streaming_decode.py:582] Cuts processed until now is 11200.
218
+ 2025-11-17 16:57:16,503 INFO [streaming_decode.py:582] Cuts processed until now is 11300.
219
+ 2025-11-17 16:57:17,451 INFO [streaming_decode.py:582] Cuts processed until now is 11400.
220
+ 2025-11-17 16:57:19,108 INFO [streaming_decode.py:582] Cuts processed until now is 11500.
221
+ 2025-11-17 16:57:20,274 INFO [streaming_decode.py:582] Cuts processed until now is 11600.
222
+ 2025-11-17 16:57:21,502 INFO [streaming_decode.py:582] Cuts processed until now is 11700.
223
+ 2025-11-17 16:57:22,526 INFO [streaming_decode.py:582] Cuts processed until now is 11800.
224
+ 2025-11-17 16:57:23,632 INFO [streaming_decode.py:582] Cuts processed until now is 11900.
225
+ 2025-11-17 16:57:24,826 INFO [streaming_decode.py:582] Cuts processed until now is 12000.
226
+ 2025-11-17 16:57:25,842 INFO [streaming_decode.py:582] Cuts processed until now is 12100.
227
+ 2025-11-17 16:57:26,857 INFO [streaming_decode.py:582] Cuts processed until now is 12200.
228
+ 2025-11-17 16:57:27,900 INFO [streaming_decode.py:582] Cuts processed until now is 12300.
229
+ 2025-11-17 16:57:29,037 INFO [streaming_decode.py:582] Cuts processed until now is 12400.
230
+ 2025-11-17 16:57:30,276 INFO [streaming_decode.py:582] Cuts processed until now is 12500.
231
+ 2025-11-17 16:57:31,470 INFO [streaming_decode.py:582] Cuts processed until now is 12600.
232
+ 2025-11-17 16:57:32,710 INFO [streaming_decode.py:582] Cuts processed until now is 12700.
233
+ 2025-11-17 16:57:33,785 INFO [streaming_decode.py:582] Cuts processed until now is 12800.
234
+ 2025-11-17 16:57:34,898 INFO [streaming_decode.py:582] Cuts processed until now is 12900.
235
+ 2025-11-17 16:57:35,934 INFO [streaming_decode.py:582] Cuts processed until now is 13000.
236
+ 2025-11-17 16:57:37,116 INFO [streaming_decode.py:582] Cuts processed until now is 13100.
237
+ 2025-11-17 16:57:38,557 INFO [streaming_decode.py:582] Cuts processed until now is 13200.
238
+ 2025-11-17 16:57:39,532 INFO [streaming_decode.py:582] Cuts processed until now is 13300.
239
+ 2025-11-17 16:57:40,931 INFO [streaming_decode.py:582] Cuts processed until now is 13400.
240
+ 2025-11-17 16:57:42,024 INFO [streaming_decode.py:582] Cuts processed until now is 13500.
241
+ 2025-11-17 16:57:43,284 INFO [streaming_decode.py:582] Cuts processed until now is 13600.
242
+ 2025-11-17 16:57:44,337 INFO [streaming_decode.py:582] Cuts processed until now is 13700.
243
+ 2025-11-17 16:57:45,316 INFO [streaming_decode.py:582] Cuts processed until now is 13800.
244
+ 2025-11-17 16:57:46,554 INFO [streaming_decode.py:582] Cuts processed until now is 13900.
245
+ 2025-11-17 16:57:47,626 INFO [streaming_decode.py:582] Cuts processed until now is 14000.
246
+ 2025-11-17 16:57:48,673 INFO [streaming_decode.py:582] Cuts processed until now is 14100.
247
+ 2025-11-17 16:57:49,459 INFO [streaming_decode.py:582] Cuts processed until now is 14200.
248
+ 2025-11-17 16:57:50,307 INFO [streaming_decode.py:582] Cuts processed until now is 14300.
249
+ 2025-11-17 16:57:51,648 INFO [streaming_decode.py:582] Cuts processed until now is 14400.
250
+ 2025-11-17 16:57:52,993 INFO [streaming_decode.py:582] Cuts processed until now is 14500.
251
+ 2025-11-17 16:57:54,212 INFO [streaming_decode.py:582] Cuts processed until now is 14600.
252
+ 2025-11-17 16:57:55,376 INFO [streaming_decode.py:582] Cuts processed until now is 14700.
253
+ 2025-11-17 16:57:56,561 INFO [streaming_decode.py:582] Cuts processed until now is 14800.
254
+ 2025-11-17 16:57:57,704 INFO [streaming_decode.py:582] Cuts processed until now is 14900.
255
+ 2025-11-17 16:57:58,679 INFO [streaming_decode.py:582] Cuts processed until now is 15000.
256
+ 2025-11-17 16:57:59,761 INFO [streaming_decode.py:582] Cuts processed until now is 15100.
257
+ 2025-11-17 16:58:00,870 INFO [streaming_decode.py:582] Cuts processed until now is 15200.
258
+ 2025-11-17 16:58:02,081 INFO [streaming_decode.py:582] Cuts processed until now is 15300.
259
+ 2025-11-17 16:58:03,129 INFO [streaming_decode.py:582] Cuts processed until now is 15400.
260
+ 2025-11-17 16:58:04,575 INFO [streaming_decode.py:582] Cuts processed until now is 15500.
261
+ 2025-11-17 16:58:05,322 INFO [streaming_decode.py:582] Cuts processed until now is 15600.
262
+ 2025-11-17 16:58:06,483 INFO [streaming_decode.py:582] Cuts processed until now is 15700.
263
+ 2025-11-17 16:58:07,324 INFO [streaming_decode.py:582] Cuts processed until now is 15800.
264
+ 2025-11-17 16:58:11,366 INFO [streaming_decode.py:618] The transcripts are stored in tmp/exp-causal-80-epoch/streaming/greedy_search/recogs-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt
265
+ 2025-11-17 16:58:11,665 INFO [utils.py:670] [test-commonvoice-greedy_search] %WER 2.56% [19960 / 778666, 3313 ins, 8477 del, 8170 sub ]
266
+ 2025-11-17 16:58:12,482 INFO [streaming_decode.py:627] Wrote detailed error stats to tmp/exp-causal-80-epoch/streaming/greedy_search/errs-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt
267
+ 2025-11-17 16:58:12,482 INFO [streaming_decode.py:641]
268
+ For test-commonvoice, WER of different settings are:
269
+ greedy_search 2.56 best for test-commonvoice
270
+
271
+ 2025-11-17 16:58:12,505 INFO [streaming_decode.py:582] Cuts processed until now is 0.
272
+ 2025-11-17 16:58:13,205 INFO [streaming_decode.py:582] Cuts processed until now is 100.
273
+ 2025-11-17 16:58:14,095 INFO [streaming_decode.py:582] Cuts processed until now is 200.
274
+ 2025-11-17 16:58:15,216 INFO [streaming_decode.py:582] Cuts processed until now is 300.
275
+ 2025-11-17 16:58:16,128 INFO [streaming_decode.py:582] Cuts processed until now is 400.
276
+ 2025-11-17 16:58:16,899 INFO [streaming_decode.py:582] Cuts processed until now is 500.
277
+ 2025-11-17 16:58:17,530 INFO [streaming_decode.py:582] Cuts processed until now is 600.
278
+ 2025-11-17 16:58:18,280 INFO [streaming_decode.py:582] Cuts processed until now is 700.
279
+ 2025-11-17 16:58:19,088 INFO [streaming_decode.py:582] Cuts processed until now is 800.
280
+ 2025-11-17 16:58:19,738 INFO [streaming_decode.py:582] Cuts processed until now is 900.
281
+ 2025-11-17 16:58:24,140 INFO [streaming_decode.py:618] The transcripts are stored in tmp/exp-causal-80-epoch/streaming/greedy_search/recogs-test-slr72-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt
282
+ 2025-11-17 16:58:24,155 INFO [utils.py:670] [test-slr72-greedy_search] %WER 1.25% [509 / 40600, 76 ins, 285 del, 148 sub ]
283
+ 2025-11-17 16:58:24,193 INFO [streaming_decode.py:627] Wrote detailed error stats to tmp/exp-causal-80-epoch/streaming/greedy_search/errs-test-slr72-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt
284
+ 2025-11-17 16:58:24,194 INFO [streaming_decode.py:641]
285
+ For test-slr72, WER of different settings are:
286
+ greedy_search 1.25 best for test-slr72
287
+
288
+ 2025-11-17 16:58:24,194 INFO [streaming_decode.py:794] Done!
streaming/greedy_search/recogs-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/greedy_search/recogs-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/greedy_search/recogs-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/greedy_search/recogs-test-slr72-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/greedy_search/wer-summary-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ settings WER
2
+ greedy_search 2.85
streaming/greedy_search/wer-summary-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ settings WER
2
+ greedy_search 2.56
streaming/greedy_search/wer-summary-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ settings WER
2
+ greedy_search 1.56
streaming/greedy_search/wer-summary-test-slr72-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ settings WER
2
+ greedy_search 1.25
streaming/modified_beam_search/errs-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/modified_beam_search/errs-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/modified_beam_search/errs-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/modified_beam_search/errs-test-slr72-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/modified_beam_search/log-decode-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model-2025-11-17-17-47-46 ADDED
@@ -0,0 +1,288 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2025-11-17 17:47:46,020 INFO [streaming_decode.py:677] Decoding started
2
+ 2025-11-17 17:47:46,020 INFO [streaming_decode.py:683] Device: cuda:0
3
+ 2025-11-17 17:47:46,020 INFO [lexicon.py:168] Loading pre-compiled data/lang_phone/Linv.pt
4
+ 2025-11-17 17:47:46,021 INFO [streaming_decode.py:691] {
5
+ "attention_decoder_attention_dim": 512,
6
+ "attention_decoder_dim": 512,
7
+ "attention_decoder_feedforward_dim": 2048,
8
+ "attention_decoder_num_heads": 8,
9
+ "attention_decoder_num_layers": 6,
10
+ "avg": 5,
11
+ "batch_idx_train": 0,
12
+ "beam": 4,
13
+ "best_train_epoch": -1,
14
+ "best_train_loss": Infinity,
15
+ "best_valid_epoch": -1,
16
+ "best_valid_loss": Infinity,
17
+ "blank_id": 0,
18
+ "bucketing_sampler": true,
19
+ "causal": true,
20
+ "chunk_size": "16",
21
+ "cnn_module_kernel": "31,31,15,15,15,31",
22
+ "concatenate_cuts": false,
23
+ "context_size": 2,
24
+ "decoder_dim": 512,
25
+ "decoding_method": "modified_beam_search",
26
+ "downsampling_factor": "1,2,4,8,4,2",
27
+ "drop_last": true,
28
+ "duration_factor": 1.0,
29
+ "enable_musan": true,
30
+ "enable_spec_aug": true,
31
+ "encoder_dim": "192,256,256,256,256,256",
32
+ "encoder_unmasked_dim": "192,192,192,192,192,192",
33
+ "env_info": {
34
+ "IP address": "127.0.1.1",
35
+ "hostname": "Bookbot-GPU1",
36
+ "icefall-git-branch": "master",
37
+ "icefall-git-date": "Tue Jan 7 16:20:44 2025",
38
+ "icefall-git-sha1": "77cd018f-dirty",
39
+ "icefall-path": "/mnt/Projects/Projects/ASR/icefall",
40
+ "k2-build-type": "Release",
41
+ "k2-git-date": "Thu Jul 25 03:46:03 2024",
42
+ "k2-git-sha1": "5735fa707f6091856d13ccd230aced6e9e64f815",
43
+ "k2-path": "/home/bookbot/miniconda3/envs/icefall/lib/python3.12/site-packages/k2/__init__.py",
44
+ "k2-version": "1.24.4",
45
+ "k2-with-cuda": true,
46
+ "lhotse-path": "/home/bookbot/miniconda3/envs/icefall/lib/python3.12/site-packages/lhotse/__init__.py",
47
+ "lhotse-version": "1.31.1.dev+git.4c43958.clean",
48
+ "python-version": "3.12",
49
+ "torch-cuda-available": true,
50
+ "torch-cuda-version": "12.4",
51
+ "torch-version": "2.4.0+cu124"
52
+ },
53
+ "epoch": 80,
54
+ "exp_dir": "tmp/exp-causal-80-epoch",
55
+ "feature_dim": 80,
56
+ "feedforward_dim": "512,768,768,768,768,768",
57
+ "gap": 1.0,
58
+ "ignore_id": -1,
59
+ "input_strategy": "PrecomputedFeatures",
60
+ "iter": 0,
61
+ "joiner_dim": 512,
62
+ "label_smoothing": 0.1,
63
+ "lang_dir": "data/lang_phone",
64
+ "left_context_frames": "128",
65
+ "log_interval": 50,
66
+ "manifest_dir": "data/fbank",
67
+ "max_contexts": 4,
68
+ "max_duration": 200.0,
69
+ "max_states": 32,
70
+ "num_active_paths": 4,
71
+ "num_buckets": 30,
72
+ "num_decode_streams": 1000,
73
+ "num_encoder_layers": "2,2,2,2,2,2",
74
+ "num_heads": "4,4,4,8,4,4",
75
+ "num_workers": 2,
76
+ "on_the_fly_feats": false,
77
+ "pos_dim": 48,
78
+ "pos_head_dim": "4",
79
+ "query_head_dim": "32",
80
+ "res_dir": "tmp/exp-causal-80-epoch/streaming/modified_beam_search",
81
+ "reset_interval": 200,
82
+ "return_cuts": true,
83
+ "shuffle": true,
84
+ "spec_aug_time_warp_factor": 80,
85
+ "subsampling_factor": 4,
86
+ "suffix": "epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model",
87
+ "unk_id": 1,
88
+ "use_attention_decoder": false,
89
+ "use_averaged_model": true,
90
+ "use_cr_ctc": false,
91
+ "use_ctc": false,
92
+ "use_transducer": true,
93
+ "valid_interval": 3000,
94
+ "value_head_dim": "12",
95
+ "vocab_size": 36,
96
+ "warm_step": 2000
97
+ }
98
+ 2025-11-17 17:47:46,021 INFO [streaming_decode.py:693] About to create model
99
+ 2025-11-17 17:47:46,191 INFO [streaming_decode.py:748] Calculating the averaged model over epoch range from 75 (excluded) to 80
100
+ 2025-11-17 17:47:47,178 INFO [streaming_decode.py:769] Number of model parameters: 22795007
101
+ 2025-11-17 17:47:47,179 INFO [multidataset.py:83] About to get Spanish Common Voice test cuts
102
+ 2025-11-17 17:47:47,179 INFO [multidataset.py:85] Loading Spanish Common Voice in lazy mode
103
+ 2025-11-17 17:47:47,179 INFO [multidataset.py:94] About to get SLR72 dataset test cuts
104
+ 2025-11-17 17:47:47,179 INFO [multidataset.py:96] Loading SLR72 dataset in lazy mode
105
+ 2025-11-17 17:47:47,297 INFO [streaming_decode.py:582] Cuts processed until now is 0.
106
+ 2025-11-17 17:47:48,204 INFO [streaming_decode.py:582] Cuts processed until now is 100.
107
+ 2025-11-17 17:47:48,849 INFO [streaming_decode.py:582] Cuts processed until now is 200.
108
+ 2025-11-17 17:47:49,378 INFO [streaming_decode.py:582] Cuts processed until now is 300.
109
+ 2025-11-17 17:47:50,260 INFO [streaming_decode.py:582] Cuts processed until now is 400.
110
+ 2025-11-17 17:47:51,169 INFO [streaming_decode.py:582] Cuts processed until now is 500.
111
+ 2025-11-17 17:47:51,796 INFO [streaming_decode.py:582] Cuts processed until now is 600.
112
+ 2025-11-17 17:47:52,471 INFO [streaming_decode.py:582] Cuts processed until now is 700.
113
+ 2025-11-17 17:47:53,249 INFO [streaming_decode.py:582] Cuts processed until now is 800.
114
+ 2025-11-17 17:47:53,951 INFO [streaming_decode.py:582] Cuts processed until now is 900.
115
+ 2025-11-17 17:48:08,183 INFO [streaming_decode.py:582] Cuts processed until now is 1000.
116
+ 2025-11-17 17:48:19,298 INFO [streaming_decode.py:582] Cuts processed until now is 1100.
117
+ 2025-11-17 17:48:23,954 INFO [streaming_decode.py:582] Cuts processed until now is 1200.
118
+ 2025-11-17 17:48:28,615 INFO [streaming_decode.py:582] Cuts processed until now is 1300.
119
+ 2025-11-17 17:48:31,441 INFO [streaming_decode.py:582] Cuts processed until now is 1400.
120
+ 2025-11-17 17:48:34,078 INFO [streaming_decode.py:582] Cuts processed until now is 1500.
121
+ 2025-11-17 17:48:38,662 INFO [streaming_decode.py:582] Cuts processed until now is 1600.
122
+ 2025-11-17 17:48:41,237 INFO [streaming_decode.py:582] Cuts processed until now is 1700.
123
+ 2025-11-17 17:48:45,142 INFO [streaming_decode.py:582] Cuts processed until now is 1800.
124
+ 2025-11-17 17:48:49,921 INFO [streaming_decode.py:582] Cuts processed until now is 1900.
125
+ 2025-11-17 17:48:56,635 INFO [streaming_decode.py:582] Cuts processed until now is 2000.
126
+ 2025-11-17 17:49:02,547 INFO [streaming_decode.py:582] Cuts processed until now is 2100.
127
+ 2025-11-17 17:49:06,603 INFO [streaming_decode.py:582] Cuts processed until now is 2200.
128
+ 2025-11-17 17:49:13,000 INFO [streaming_decode.py:582] Cuts processed until now is 2300.
129
+ 2025-11-17 17:49:15,476 INFO [streaming_decode.py:582] Cuts processed until now is 2400.
130
+ 2025-11-17 17:49:19,817 INFO [streaming_decode.py:582] Cuts processed until now is 2500.
131
+ 2025-11-17 17:49:23,793 INFO [streaming_decode.py:582] Cuts processed until now is 2600.
132
+ 2025-11-17 17:49:28,027 INFO [streaming_decode.py:582] Cuts processed until now is 2700.
133
+ 2025-11-17 17:49:31,635 INFO [streaming_decode.py:582] Cuts processed until now is 2800.
134
+ 2025-11-17 17:49:36,356 INFO [streaming_decode.py:582] Cuts processed until now is 2900.
135
+ 2025-11-17 17:49:40,336 INFO [streaming_decode.py:582] Cuts processed until now is 3000.
136
+ 2025-11-17 17:49:46,136 INFO [streaming_decode.py:582] Cuts processed until now is 3100.
137
+ 2025-11-17 17:49:49,999 INFO [streaming_decode.py:582] Cuts processed until now is 3200.
138
+ 2025-11-17 17:49:54,307 INFO [streaming_decode.py:582] Cuts processed until now is 3300.
139
+ 2025-11-17 17:49:58,402 INFO [streaming_decode.py:582] Cuts processed until now is 3400.
140
+ 2025-11-17 17:50:02,912 INFO [streaming_decode.py:582] Cuts processed until now is 3500.
141
+ 2025-11-17 17:50:06,877 INFO [streaming_decode.py:582] Cuts processed until now is 3600.
142
+ 2025-11-17 17:50:11,667 INFO [streaming_decode.py:582] Cuts processed until now is 3700.
143
+ 2025-11-17 17:50:15,887 INFO [streaming_decode.py:582] Cuts processed until now is 3800.
144
+ 2025-11-17 17:50:21,784 INFO [streaming_decode.py:582] Cuts processed until now is 3900.
145
+ 2025-11-17 17:50:26,069 INFO [streaming_decode.py:582] Cuts processed until now is 4000.
146
+ 2025-11-17 17:50:30,457 INFO [streaming_decode.py:582] Cuts processed until now is 4100.
147
+ 2025-11-17 17:50:34,852 INFO [streaming_decode.py:582] Cuts processed until now is 4200.
148
+ 2025-11-17 17:50:39,479 INFO [streaming_decode.py:582] Cuts processed until now is 4300.
149
+ 2025-11-17 17:50:44,844 INFO [streaming_decode.py:582] Cuts processed until now is 4400.
150
+ 2025-11-17 17:50:49,011 INFO [streaming_decode.py:582] Cuts processed until now is 4500.
151
+ 2025-11-17 17:50:53,750 INFO [streaming_decode.py:582] Cuts processed until now is 4600.
152
+ 2025-11-17 17:50:58,572 INFO [streaming_decode.py:582] Cuts processed until now is 4700.
153
+ 2025-11-17 17:51:02,443 INFO [streaming_decode.py:582] Cuts processed until now is 4800.
154
+ 2025-11-17 17:51:06,345 INFO [streaming_decode.py:582] Cuts processed until now is 4900.
155
+ 2025-11-17 17:51:11,539 INFO [streaming_decode.py:582] Cuts processed until now is 5000.
156
+ 2025-11-17 17:51:15,520 INFO [streaming_decode.py:582] Cuts processed until now is 5100.
157
+ 2025-11-17 17:51:22,321 INFO [streaming_decode.py:582] Cuts processed until now is 5200.
158
+ 2025-11-17 17:51:26,866 INFO [streaming_decode.py:582] Cuts processed until now is 5300.
159
+ 2025-11-17 17:51:31,275 INFO [streaming_decode.py:582] Cuts processed until now is 5400.
160
+ 2025-11-17 17:51:36,574 INFO [streaming_decode.py:582] Cuts processed until now is 5500.
161
+ 2025-11-17 17:51:39,016 INFO [streaming_decode.py:582] Cuts processed until now is 5600.
162
+ 2025-11-17 17:51:44,874 INFO [streaming_decode.py:582] Cuts processed until now is 5700.
163
+ 2025-11-17 17:51:48,880 INFO [streaming_decode.py:582] Cuts processed until now is 5800.
164
+ 2025-11-17 17:51:53,855 INFO [streaming_decode.py:582] Cuts processed until now is 5900.
165
+ 2025-11-17 17:51:58,737 INFO [streaming_decode.py:582] Cuts processed until now is 6000.
166
+ 2025-11-17 17:52:05,133 INFO [streaming_decode.py:582] Cuts processed until now is 6100.
167
+ 2025-11-17 17:52:09,915 INFO [streaming_decode.py:582] Cuts processed until now is 6200.
168
+ 2025-11-17 17:52:14,576 INFO [streaming_decode.py:582] Cuts processed until now is 6300.
169
+ 2025-11-17 17:52:19,657 INFO [streaming_decode.py:582] Cuts processed until now is 6400.
170
+ 2025-11-17 17:52:24,068 INFO [streaming_decode.py:582] Cuts processed until now is 6500.
171
+ 2025-11-17 17:52:28,501 INFO [streaming_decode.py:582] Cuts processed until now is 6600.
172
+ 2025-11-17 17:52:33,254 INFO [streaming_decode.py:582] Cuts processed until now is 6700.
173
+ 2025-11-17 17:52:38,225 INFO [streaming_decode.py:582] Cuts processed until now is 6800.
174
+ 2025-11-17 17:52:42,574 INFO [streaming_decode.py:582] Cuts processed until now is 6900.
175
+ 2025-11-17 17:52:47,499 INFO [streaming_decode.py:582] Cuts processed until now is 7000.
176
+ 2025-11-17 17:52:52,545 INFO [streaming_decode.py:582] Cuts processed until now is 7100.
177
+ 2025-11-17 17:52:59,145 INFO [streaming_decode.py:582] Cuts processed until now is 7200.
178
+ 2025-11-17 17:53:03,519 INFO [streaming_decode.py:582] Cuts processed until now is 7300.
179
+ 2025-11-17 17:53:08,328 INFO [streaming_decode.py:582] Cuts processed until now is 7400.
180
+ 2025-11-17 17:53:12,928 INFO [streaming_decode.py:582] Cuts processed until now is 7500.
181
+ 2025-11-17 17:53:17,828 INFO [streaming_decode.py:582] Cuts processed until now is 7600.
182
+ 2025-11-17 17:53:22,546 INFO [streaming_decode.py:582] Cuts processed until now is 7700.
183
+ 2025-11-17 17:53:27,352 INFO [streaming_decode.py:582] Cuts processed until now is 7800.
184
+ 2025-11-17 17:53:32,576 INFO [streaming_decode.py:582] Cuts processed until now is 7900.
185
+ 2025-11-17 17:53:36,703 INFO [streaming_decode.py:582] Cuts processed until now is 8000.
186
+ 2025-11-17 17:53:41,697 INFO [streaming_decode.py:582] Cuts processed until now is 8100.
187
+ 2025-11-17 17:53:46,811 INFO [streaming_decode.py:582] Cuts processed until now is 8200.
188
+ 2025-11-17 17:53:51,628 INFO [streaming_decode.py:582] Cuts processed until now is 8300.
189
+ 2025-11-17 17:53:56,008 INFO [streaming_decode.py:582] Cuts processed until now is 8400.
190
+ 2025-11-17 17:54:01,764 INFO [streaming_decode.py:582] Cuts processed until now is 8500.
191
+ 2025-11-17 17:54:06,050 INFO [streaming_decode.py:582] Cuts processed until now is 8600.
192
+ 2025-11-17 17:54:10,087 INFO [streaming_decode.py:582] Cuts processed until now is 8700.
193
+ 2025-11-17 17:54:15,038 INFO [streaming_decode.py:582] Cuts processed until now is 8800.
194
+ 2025-11-17 17:54:19,498 INFO [streaming_decode.py:582] Cuts processed until now is 8900.
195
+ 2025-11-17 17:54:24,300 INFO [streaming_decode.py:582] Cuts processed until now is 9000.
196
+ 2025-11-17 17:54:29,241 INFO [streaming_decode.py:582] Cuts processed until now is 9100.
197
+ 2025-11-17 17:54:34,486 INFO [streaming_decode.py:582] Cuts processed until now is 9200.
198
+ 2025-11-17 17:54:38,174 INFO [streaming_decode.py:582] Cuts processed until now is 9300.
199
+ 2025-11-17 17:54:42,523 INFO [streaming_decode.py:582] Cuts processed until now is 9400.
200
+ 2025-11-17 17:54:46,867 INFO [streaming_decode.py:582] Cuts processed until now is 9500.
201
+ 2025-11-17 17:54:53,109 INFO [streaming_decode.py:582] Cuts processed until now is 9600.
202
+ 2025-11-17 17:54:57,693 INFO [streaming_decode.py:582] Cuts processed until now is 9700.
203
+ 2025-11-17 17:55:02,500 INFO [streaming_decode.py:582] Cuts processed until now is 9800.
204
+ 2025-11-17 17:55:06,742 INFO [streaming_decode.py:582] Cuts processed until now is 9900.
205
+ 2025-11-17 17:55:10,645 INFO [streaming_decode.py:582] Cuts processed until now is 10000.
206
+ 2025-11-17 17:55:15,099 INFO [streaming_decode.py:582] Cuts processed until now is 10100.
207
+ 2025-11-17 17:55:18,819 INFO [streaming_decode.py:582] Cuts processed until now is 10200.
208
+ 2025-11-17 17:55:23,397 INFO [streaming_decode.py:582] Cuts processed until now is 10300.
209
+ 2025-11-17 17:55:27,556 INFO [streaming_decode.py:582] Cuts processed until now is 10400.
210
+ 2025-11-17 17:55:33,620 INFO [streaming_decode.py:582] Cuts processed until now is 10500.
211
+ 2025-11-17 17:55:36,080 INFO [streaming_decode.py:582] Cuts processed until now is 10600.
212
+ 2025-11-17 17:55:40,697 INFO [streaming_decode.py:582] Cuts processed until now is 10700.
213
+ 2025-11-17 17:55:45,142 INFO [streaming_decode.py:582] Cuts processed until now is 10800.
214
+ 2025-11-17 17:55:49,812 INFO [streaming_decode.py:582] Cuts processed until now is 10900.
215
+ 2025-11-17 17:55:55,807 INFO [streaming_decode.py:582] Cuts processed until now is 11000.
216
+ 2025-11-17 17:55:59,810 INFO [streaming_decode.py:582] Cuts processed until now is 11100.
217
+ 2025-11-17 17:56:04,860 INFO [streaming_decode.py:582] Cuts processed until now is 11200.
218
+ 2025-11-17 17:56:09,247 INFO [streaming_decode.py:582] Cuts processed until now is 11300.
219
+ 2025-11-17 17:56:13,858 INFO [streaming_decode.py:582] Cuts processed until now is 11400.
220
+ 2025-11-17 17:56:19,697 INFO [streaming_decode.py:582] Cuts processed until now is 11500.
221
+ 2025-11-17 17:56:24,043 INFO [streaming_decode.py:582] Cuts processed until now is 11600.
222
+ 2025-11-17 17:56:28,497 INFO [streaming_decode.py:582] Cuts processed until now is 11700.
223
+ 2025-11-17 17:56:33,593 INFO [streaming_decode.py:582] Cuts processed until now is 11800.
224
+ 2025-11-17 17:56:36,512 INFO [streaming_decode.py:582] Cuts processed until now is 11900.
225
+ 2025-11-17 17:56:43,483 INFO [streaming_decode.py:582] Cuts processed until now is 12000.
226
+ 2025-11-17 17:56:48,213 INFO [streaming_decode.py:582] Cuts processed until now is 12100.
227
+ 2025-11-17 17:56:52,337 INFO [streaming_decode.py:582] Cuts processed until now is 12200.
228
+ 2025-11-17 17:56:56,652 INFO [streaming_decode.py:582] Cuts processed until now is 12300.
229
+ 2025-11-17 17:57:01,410 INFO [streaming_decode.py:582] Cuts processed until now is 12400.
230
+ 2025-11-17 17:57:05,745 INFO [streaming_decode.py:582] Cuts processed until now is 12500.
231
+ 2025-11-17 17:57:12,132 INFO [streaming_decode.py:582] Cuts processed until now is 12600.
232
+ 2025-11-17 17:57:14,949 INFO [streaming_decode.py:582] Cuts processed until now is 12700.
233
+ 2025-11-17 17:57:19,510 INFO [streaming_decode.py:582] Cuts processed until now is 12800.
234
+ 2025-11-17 17:57:24,451 INFO [streaming_decode.py:582] Cuts processed until now is 12900.
235
+ 2025-11-17 17:57:30,686 INFO [streaming_decode.py:582] Cuts processed until now is 13000.
236
+ 2025-11-17 17:57:35,322 INFO [streaming_decode.py:582] Cuts processed until now is 13100.
237
+ 2025-11-17 17:57:39,876 INFO [streaming_decode.py:582] Cuts processed until now is 13200.
238
+ 2025-11-17 17:57:44,605 INFO [streaming_decode.py:582] Cuts processed until now is 13300.
239
+ 2025-11-17 17:57:48,967 INFO [streaming_decode.py:582] Cuts processed until now is 13400.
240
+ 2025-11-17 17:57:53,130 INFO [streaming_decode.py:582] Cuts processed until now is 13500.
241
+ 2025-11-17 17:57:57,581 INFO [streaming_decode.py:582] Cuts processed until now is 13600.
242
+ 2025-11-17 17:58:01,675 INFO [streaming_decode.py:582] Cuts processed until now is 13700.
243
+ 2025-11-17 17:58:06,175 INFO [streaming_decode.py:582] Cuts processed until now is 13800.
244
+ 2025-11-17 17:58:11,397 INFO [streaming_decode.py:582] Cuts processed until now is 13900.
245
+ 2025-11-17 17:58:16,197 INFO [streaming_decode.py:582] Cuts processed until now is 14000.
246
+ 2025-11-17 17:58:21,244 INFO [streaming_decode.py:582] Cuts processed until now is 14100.
247
+ 2025-11-17 17:58:27,800 INFO [streaming_decode.py:582] Cuts processed until now is 14200.
248
+ 2025-11-17 17:58:32,332 INFO [streaming_decode.py:582] Cuts processed until now is 14300.
249
+ 2025-11-17 17:58:36,744 INFO [streaming_decode.py:582] Cuts processed until now is 14400.
250
+ 2025-11-17 17:58:41,866 INFO [streaming_decode.py:582] Cuts processed until now is 14500.
251
+ 2025-11-17 17:58:46,025 INFO [streaming_decode.py:582] Cuts processed until now is 14600.
252
+ 2025-11-17 17:58:52,898 INFO [streaming_decode.py:582] Cuts processed until now is 14700.
253
+ 2025-11-17 17:58:57,657 INFO [streaming_decode.py:582] Cuts processed until now is 14800.
254
+ 2025-11-17 17:59:01,506 INFO [streaming_decode.py:582] Cuts processed until now is 14900.
255
+ 2025-11-17 17:59:05,937 INFO [streaming_decode.py:582] Cuts processed until now is 15000.
256
+ 2025-11-17 17:59:10,694 INFO [streaming_decode.py:582] Cuts processed until now is 15100.
257
+ 2025-11-17 17:59:15,125 INFO [streaming_decode.py:582] Cuts processed until now is 15200.
258
+ 2025-11-17 17:59:19,517 INFO [streaming_decode.py:582] Cuts processed until now is 15300.
259
+ 2025-11-17 17:59:24,369 INFO [streaming_decode.py:582] Cuts processed until now is 15400.
260
+ 2025-11-17 17:59:28,994 INFO [streaming_decode.py:582] Cuts processed until now is 15500.
261
+ 2025-11-17 17:59:33,692 INFO [streaming_decode.py:582] Cuts processed until now is 15600.
262
+ 2025-11-17 17:59:38,506 INFO [streaming_decode.py:582] Cuts processed until now is 15700.
263
+ 2025-11-17 17:59:42,738 INFO [streaming_decode.py:582] Cuts processed until now is 15800.
264
+ 2025-11-17 18:00:10,048 INFO [streaming_decode.py:618] The transcripts are stored in tmp/exp-causal-80-epoch/streaming/modified_beam_search/recogs-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt
265
+ 2025-11-17 18:00:10,405 INFO [utils.py:670] [test-commonvoice-num_active_paths_4] %WER 2.71% [21085 / 778666, 3990 ins, 8104 del, 8991 sub ]
266
+ 2025-11-17 18:00:11,223 INFO [streaming_decode.py:627] Wrote detailed error stats to tmp/exp-causal-80-epoch/streaming/modified_beam_search/errs-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt
267
+ 2025-11-17 18:00:11,223 INFO [streaming_decode.py:641]
268
+ For test-commonvoice, WER of different settings are:
269
+ num_active_paths_4 2.71 best for test-commonvoice
270
+
271
+ 2025-11-17 18:00:11,252 INFO [streaming_decode.py:582] Cuts processed until now is 0.
272
+ 2025-11-17 18:00:11,986 INFO [streaming_decode.py:582] Cuts processed until now is 100.
273
+ 2025-11-17 18:00:12,934 INFO [streaming_decode.py:582] Cuts processed until now is 200.
274
+ 2025-11-17 18:00:13,818 INFO [streaming_decode.py:582] Cuts processed until now is 300.
275
+ 2025-11-17 18:00:14,568 INFO [streaming_decode.py:582] Cuts processed until now is 400.
276
+ 2025-11-17 18:00:15,431 INFO [streaming_decode.py:582] Cuts processed until now is 500.
277
+ 2025-11-17 18:00:16,058 INFO [streaming_decode.py:582] Cuts processed until now is 600.
278
+ 2025-11-17 18:00:16,525 INFO [streaming_decode.py:582] Cuts processed until now is 700.
279
+ 2025-11-17 18:00:17,085 INFO [streaming_decode.py:582] Cuts processed until now is 800.
280
+ 2025-11-17 18:00:17,910 INFO [streaming_decode.py:582] Cuts processed until now is 900.
281
+ 2025-11-17 18:00:53,173 INFO [streaming_decode.py:618] The transcripts are stored in tmp/exp-causal-80-epoch/streaming/modified_beam_search/recogs-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt
282
+ 2025-11-17 18:00:53,192 INFO [utils.py:670] [test-slr72-num_active_paths_4] %WER 1.47% [597 / 40600, 96 ins, 327 del, 174 sub ]
283
+ 2025-11-17 18:00:53,245 INFO [streaming_decode.py:627] Wrote detailed error stats to tmp/exp-causal-80-epoch/streaming/modified_beam_search/errs-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt
284
+ 2025-11-17 18:00:53,245 INFO [streaming_decode.py:641]
285
+ For test-slr72, WER of different settings are:
286
+ num_active_paths_4 1.47 best for test-slr72
287
+
288
+ 2025-11-17 18:00:53,245 INFO [streaming_decode.py:794] Done!
streaming/modified_beam_search/log-decode-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model-2025-11-17-16-49-52 ADDED
@@ -0,0 +1,114 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2025-11-17 16:49:52,151 INFO [streaming_decode.py:677] Decoding started
2
+ 2025-11-17 16:49:52,151 INFO [streaming_decode.py:683] Device: cuda:0
3
+ 2025-11-17 16:49:52,152 INFO [lexicon.py:168] Loading pre-compiled data/lang_phone/Linv.pt
4
+ 2025-11-17 16:49:52,153 INFO [streaming_decode.py:691] {
5
+ "attention_decoder_attention_dim": 512,
6
+ "attention_decoder_dim": 512,
7
+ "attention_decoder_feedforward_dim": 2048,
8
+ "attention_decoder_num_heads": 8,
9
+ "attention_decoder_num_layers": 6,
10
+ "avg": 5,
11
+ "batch_idx_train": 0,
12
+ "beam": 4,
13
+ "best_train_epoch": -1,
14
+ "best_train_loss": Infinity,
15
+ "best_valid_epoch": -1,
16
+ "best_valid_loss": Infinity,
17
+ "blank_id": 0,
18
+ "bucketing_sampler": true,
19
+ "causal": true,
20
+ "chunk_size": "32",
21
+ "cnn_module_kernel": "31,31,15,15,15,31",
22
+ "concatenate_cuts": false,
23
+ "context_size": 2,
24
+ "decoder_dim": 512,
25
+ "decoding_method": "modified_beam_search",
26
+ "downsampling_factor": "1,2,4,8,4,2",
27
+ "drop_last": true,
28
+ "duration_factor": 1.0,
29
+ "enable_musan": true,
30
+ "enable_spec_aug": true,
31
+ "encoder_dim": "192,256,256,256,256,256",
32
+ "encoder_unmasked_dim": "192,192,192,192,192,192",
33
+ "env_info": {
34
+ "IP address": "127.0.1.1",
35
+ "hostname": "Bookbot-GPU1",
36
+ "icefall-git-branch": "master",
37
+ "icefall-git-date": "Tue Jan 7 16:20:44 2025",
38
+ "icefall-git-sha1": "77cd018f-dirty",
39
+ "icefall-path": "/mnt/Projects/Projects/ASR/icefall",
40
+ "k2-build-type": "Release",
41
+ "k2-git-date": "Thu Jul 25 03:46:03 2024",
42
+ "k2-git-sha1": "5735fa707f6091856d13ccd230aced6e9e64f815",
43
+ "k2-path": "/home/bookbot/miniconda3/envs/icefall/lib/python3.12/site-packages/k2/__init__.py",
44
+ "k2-version": "1.24.4",
45
+ "k2-with-cuda": true,
46
+ "lhotse-path": "/home/bookbot/miniconda3/envs/icefall/lib/python3.12/site-packages/lhotse/__init__.py",
47
+ "lhotse-version": "1.31.1.dev+git.4c43958.clean",
48
+ "python-version": "3.12",
49
+ "torch-cuda-available": true,
50
+ "torch-cuda-version": "12.4",
51
+ "torch-version": "2.4.0+cu124"
52
+ },
53
+ "epoch": 80,
54
+ "exp_dir": "tmp/exp-causal-80-epoch",
55
+ "feature_dim": 80,
56
+ "feedforward_dim": "512,768,768,768,768,768",
57
+ "gap": 1.0,
58
+ "ignore_id": -1,
59
+ "input_strategy": "PrecomputedFeatures",
60
+ "iter": 0,
61
+ "joiner_dim": 512,
62
+ "label_smoothing": 0.1,
63
+ "lang_dir": "data/lang_phone",
64
+ "left_context_frames": "128",
65
+ "log_interval": 50,
66
+ "manifest_dir": "data/fbank",
67
+ "max_contexts": 4,
68
+ "max_duration": 200.0,
69
+ "max_states": 32,
70
+ "num_active_paths": 4,
71
+ "num_buckets": 30,
72
+ "num_decode_streams": 1000,
73
+ "num_encoder_layers": "2,2,2,2,2,2",
74
+ "num_heads": "4,4,4,8,4,4",
75
+ "num_workers": 2,
76
+ "on_the_fly_feats": false,
77
+ "pos_dim": 48,
78
+ "pos_head_dim": "4",
79
+ "query_head_dim": "32",
80
+ "res_dir": "tmp/exp-causal-80-epoch/streaming/modified_beam_search",
81
+ "reset_interval": 200,
82
+ "return_cuts": true,
83
+ "shuffle": true,
84
+ "spec_aug_time_warp_factor": 80,
85
+ "subsampling_factor": 4,
86
+ "suffix": "epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model",
87
+ "unk_id": 1,
88
+ "use_attention_decoder": false,
89
+ "use_averaged_model": true,
90
+ "use_cr_ctc": false,
91
+ "use_ctc": false,
92
+ "use_transducer": true,
93
+ "valid_interval": 3000,
94
+ "value_head_dim": "12",
95
+ "vocab_size": 36,
96
+ "warm_step": 2000
97
+ }
98
+ 2025-11-17 16:49:52,153 INFO [streaming_decode.py:693] About to create model
99
+ 2025-11-17 16:49:52,433 INFO [streaming_decode.py:748] Calculating the averaged model over epoch range from 75 (excluded) to 80
100
+ 2025-11-17 16:49:54,143 INFO [streaming_decode.py:769] Number of model parameters: 22795007
101
+ 2025-11-17 16:49:54,143 INFO [multidataset.py:83] About to get Spanish Common Voice test cuts
102
+ 2025-11-17 16:49:54,143 INFO [multidataset.py:85] Loading Spanish Common Voice in lazy mode
103
+ 2025-11-17 16:49:54,144 INFO [multidataset.py:94] About to get SLR72 dataset test cuts
104
+ 2025-11-17 16:49:54,144 INFO [multidataset.py:96] Loading SLR72 dataset in lazy mode
105
+ 2025-11-17 16:49:54,320 INFO [streaming_decode.py:582] Cuts processed until now is 0.
106
+ 2025-11-17 16:49:56,681 INFO [streaming_decode.py:582] Cuts processed until now is 100.
107
+ 2025-11-17 16:49:58,404 INFO [streaming_decode.py:582] Cuts processed until now is 200.
108
+ 2025-11-17 16:50:00,846 INFO [streaming_decode.py:582] Cuts processed until now is 300.
109
+ 2025-11-17 16:50:02,614 INFO [streaming_decode.py:582] Cuts processed until now is 400.
110
+ 2025-11-17 16:50:05,448 INFO [streaming_decode.py:582] Cuts processed until now is 500.
111
+ 2025-11-17 16:50:07,663 INFO [streaming_decode.py:582] Cuts processed until now is 600.
112
+ 2025-11-17 16:50:09,718 INFO [streaming_decode.py:582] Cuts processed until now is 700.
113
+ 2025-11-17 16:50:11,793 INFO [streaming_decode.py:582] Cuts processed until now is 800.
114
+ 2025-11-17 16:50:13,844 INFO [streaming_decode.py:582] Cuts processed until now is 900.
streaming/modified_beam_search/log-decode-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model-2025-11-17-17-03-31 ADDED
@@ -0,0 +1,288 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2025-11-17 17:03:31,357 INFO [streaming_decode.py:677] Decoding started
2
+ 2025-11-17 17:03:31,357 INFO [streaming_decode.py:683] Device: cuda:0
3
+ 2025-11-17 17:03:31,358 INFO [lexicon.py:168] Loading pre-compiled data/lang_phone/Linv.pt
4
+ 2025-11-17 17:03:31,359 INFO [streaming_decode.py:691] {
5
+ "attention_decoder_attention_dim": 512,
6
+ "attention_decoder_dim": 512,
7
+ "attention_decoder_feedforward_dim": 2048,
8
+ "attention_decoder_num_heads": 8,
9
+ "attention_decoder_num_layers": 6,
10
+ "avg": 5,
11
+ "batch_idx_train": 0,
12
+ "beam": 4,
13
+ "best_train_epoch": -1,
14
+ "best_train_loss": Infinity,
15
+ "best_valid_epoch": -1,
16
+ "best_valid_loss": Infinity,
17
+ "blank_id": 0,
18
+ "bucketing_sampler": true,
19
+ "causal": true,
20
+ "chunk_size": "32",
21
+ "cnn_module_kernel": "31,31,15,15,15,31",
22
+ "concatenate_cuts": false,
23
+ "context_size": 2,
24
+ "decoder_dim": 512,
25
+ "decoding_method": "modified_beam_search",
26
+ "downsampling_factor": "1,2,4,8,4,2",
27
+ "drop_last": true,
28
+ "duration_factor": 1.0,
29
+ "enable_musan": true,
30
+ "enable_spec_aug": true,
31
+ "encoder_dim": "192,256,256,256,256,256",
32
+ "encoder_unmasked_dim": "192,192,192,192,192,192",
33
+ "env_info": {
34
+ "IP address": "127.0.1.1",
35
+ "hostname": "Bookbot-GPU1",
36
+ "icefall-git-branch": "master",
37
+ "icefall-git-date": "Tue Jan 7 16:20:44 2025",
38
+ "icefall-git-sha1": "77cd018f-dirty",
39
+ "icefall-path": "/mnt/Projects/Projects/ASR/icefall",
40
+ "k2-build-type": "Release",
41
+ "k2-git-date": "Thu Jul 25 03:46:03 2024",
42
+ "k2-git-sha1": "5735fa707f6091856d13ccd230aced6e9e64f815",
43
+ "k2-path": "/home/bookbot/miniconda3/envs/icefall/lib/python3.12/site-packages/k2/__init__.py",
44
+ "k2-version": "1.24.4",
45
+ "k2-with-cuda": true,
46
+ "lhotse-path": "/home/bookbot/miniconda3/envs/icefall/lib/python3.12/site-packages/lhotse/__init__.py",
47
+ "lhotse-version": "1.31.1.dev+git.4c43958.clean",
48
+ "python-version": "3.12",
49
+ "torch-cuda-available": true,
50
+ "torch-cuda-version": "12.4",
51
+ "torch-version": "2.4.0+cu124"
52
+ },
53
+ "epoch": 80,
54
+ "exp_dir": "tmp/exp-causal-80-epoch",
55
+ "feature_dim": 80,
56
+ "feedforward_dim": "512,768,768,768,768,768",
57
+ "gap": 1.0,
58
+ "ignore_id": -1,
59
+ "input_strategy": "PrecomputedFeatures",
60
+ "iter": 0,
61
+ "joiner_dim": 512,
62
+ "label_smoothing": 0.1,
63
+ "lang_dir": "data/lang_phone",
64
+ "left_context_frames": "128",
65
+ "log_interval": 50,
66
+ "manifest_dir": "data/fbank",
67
+ "max_contexts": 4,
68
+ "max_duration": 200.0,
69
+ "max_states": 32,
70
+ "num_active_paths": 4,
71
+ "num_buckets": 30,
72
+ "num_decode_streams": 1000,
73
+ "num_encoder_layers": "2,2,2,2,2,2",
74
+ "num_heads": "4,4,4,8,4,4",
75
+ "num_workers": 2,
76
+ "on_the_fly_feats": false,
77
+ "pos_dim": 48,
78
+ "pos_head_dim": "4",
79
+ "query_head_dim": "32",
80
+ "res_dir": "tmp/exp-causal-80-epoch/streaming/modified_beam_search",
81
+ "reset_interval": 200,
82
+ "return_cuts": true,
83
+ "shuffle": true,
84
+ "spec_aug_time_warp_factor": 80,
85
+ "subsampling_factor": 4,
86
+ "suffix": "epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model",
87
+ "unk_id": 1,
88
+ "use_attention_decoder": false,
89
+ "use_averaged_model": true,
90
+ "use_cr_ctc": false,
91
+ "use_ctc": false,
92
+ "use_transducer": true,
93
+ "valid_interval": 3000,
94
+ "value_head_dim": "12",
95
+ "vocab_size": 36,
96
+ "warm_step": 2000
97
+ }
98
+ 2025-11-17 17:03:31,359 INFO [streaming_decode.py:693] About to create model
99
+ 2025-11-17 17:03:31,533 INFO [streaming_decode.py:748] Calculating the averaged model over epoch range from 75 (excluded) to 80
100
+ 2025-11-17 17:03:32,210 INFO [streaming_decode.py:769] Number of model parameters: 22795007
101
+ 2025-11-17 17:03:32,210 INFO [multidataset.py:83] About to get Spanish Common Voice test cuts
102
+ 2025-11-17 17:03:32,210 INFO [multidataset.py:85] Loading Spanish Common Voice in lazy mode
103
+ 2025-11-17 17:03:32,211 INFO [multidataset.py:94] About to get SLR72 dataset test cuts
104
+ 2025-11-17 17:03:32,211 INFO [multidataset.py:96] Loading SLR72 dataset in lazy mode
105
+ 2025-11-17 17:03:32,303 INFO [streaming_decode.py:582] Cuts processed until now is 0.
106
+ 2025-11-17 17:03:32,971 INFO [streaming_decode.py:582] Cuts processed until now is 100.
107
+ 2025-11-17 17:03:33,924 INFO [streaming_decode.py:582] Cuts processed until now is 200.
108
+ 2025-11-17 17:03:34,729 INFO [streaming_decode.py:582] Cuts processed until now is 300.
109
+ 2025-11-17 17:03:35,544 INFO [streaming_decode.py:582] Cuts processed until now is 400.
110
+ 2025-11-17 17:03:36,548 INFO [streaming_decode.py:582] Cuts processed until now is 500.
111
+ 2025-11-17 17:03:37,403 INFO [streaming_decode.py:582] Cuts processed until now is 600.
112
+ 2025-11-17 17:03:38,383 INFO [streaming_decode.py:582] Cuts processed until now is 700.
113
+ 2025-11-17 17:03:39,246 INFO [streaming_decode.py:582] Cuts processed until now is 800.
114
+ 2025-11-17 17:03:40,203 INFO [streaming_decode.py:582] Cuts processed until now is 900.
115
+ 2025-11-17 17:03:55,119 INFO [streaming_decode.py:582] Cuts processed until now is 1000.
116
+ 2025-11-17 17:04:06,879 INFO [streaming_decode.py:582] Cuts processed until now is 1100.
117
+ 2025-11-17 17:04:11,466 INFO [streaming_decode.py:582] Cuts processed until now is 1200.
118
+ 2025-11-17 17:04:16,142 INFO [streaming_decode.py:582] Cuts processed until now is 1300.
119
+ 2025-11-17 17:04:20,396 INFO [streaming_decode.py:582] Cuts processed until now is 1400.
120
+ 2025-11-17 17:04:21,258 INFO [streaming_decode.py:582] Cuts processed until now is 1500.
121
+ 2025-11-17 17:04:25,759 INFO [streaming_decode.py:582] Cuts processed until now is 1600.
122
+ 2025-11-17 17:04:29,876 INFO [streaming_decode.py:582] Cuts processed until now is 1700.
123
+ 2025-11-17 17:04:33,900 INFO [streaming_decode.py:582] Cuts processed until now is 1800.
124
+ 2025-11-17 17:04:38,068 INFO [streaming_decode.py:582] Cuts processed until now is 1900.
125
+ 2025-11-17 17:04:46,017 INFO [streaming_decode.py:582] Cuts processed until now is 2000.
126
+ 2025-11-17 17:04:50,241 INFO [streaming_decode.py:582] Cuts processed until now is 2100.
127
+ 2025-11-17 17:04:54,726 INFO [streaming_decode.py:582] Cuts processed until now is 2200.
128
+ 2025-11-17 17:04:59,550 INFO [streaming_decode.py:582] Cuts processed until now is 2300.
129
+ 2025-11-17 17:05:04,184 INFO [streaming_decode.py:582] Cuts processed until now is 2400.
130
+ 2025-11-17 17:05:08,360 INFO [streaming_decode.py:582] Cuts processed until now is 2500.
131
+ 2025-11-17 17:05:12,873 INFO [streaming_decode.py:582] Cuts processed until now is 2600.
132
+ 2025-11-17 17:05:17,167 INFO [streaming_decode.py:582] Cuts processed until now is 2700.
133
+ 2025-11-17 17:05:21,087 INFO [streaming_decode.py:582] Cuts processed until now is 2800.
134
+ 2025-11-17 17:05:25,557 INFO [streaming_decode.py:582] Cuts processed until now is 2900.
135
+ 2025-11-17 17:05:29,399 INFO [streaming_decode.py:582] Cuts processed until now is 3000.
136
+ 2025-11-17 17:05:36,702 INFO [streaming_decode.py:582] Cuts processed until now is 3100.
137
+ 2025-11-17 17:05:40,999 INFO [streaming_decode.py:582] Cuts processed until now is 3200.
138
+ 2025-11-17 17:05:45,678 INFO [streaming_decode.py:582] Cuts processed until now is 3300.
139
+ 2025-11-17 17:05:50,044 INFO [streaming_decode.py:582] Cuts processed until now is 3400.
140
+ 2025-11-17 17:05:54,318 INFO [streaming_decode.py:582] Cuts processed until now is 3500.
141
+ 2025-11-17 17:05:58,714 INFO [streaming_decode.py:582] Cuts processed until now is 3600.
142
+ 2025-11-17 17:06:02,744 INFO [streaming_decode.py:582] Cuts processed until now is 3700.
143
+ 2025-11-17 17:06:06,920 INFO [streaming_decode.py:582] Cuts processed until now is 3800.
144
+ 2025-11-17 17:06:11,692 INFO [streaming_decode.py:582] Cuts processed until now is 3900.
145
+ 2025-11-17 17:06:16,299 INFO [streaming_decode.py:582] Cuts processed until now is 4000.
146
+ 2025-11-17 17:06:20,673 INFO [streaming_decode.py:582] Cuts processed until now is 4100.
147
+ 2025-11-17 17:06:25,238 INFO [streaming_decode.py:582] Cuts processed until now is 4200.
148
+ 2025-11-17 17:06:30,091 INFO [streaming_decode.py:582] Cuts processed until now is 4300.
149
+ 2025-11-17 17:06:34,647 INFO [streaming_decode.py:582] Cuts processed until now is 4400.
150
+ 2025-11-17 17:06:38,941 INFO [streaming_decode.py:582] Cuts processed until now is 4500.
151
+ 2025-11-17 17:06:43,251 INFO [streaming_decode.py:582] Cuts processed until now is 4600.
152
+ 2025-11-17 17:06:48,202 INFO [streaming_decode.py:582] Cuts processed until now is 4700.
153
+ 2025-11-17 17:06:55,400 INFO [streaming_decode.py:582] Cuts processed until now is 4800.
154
+ 2025-11-17 17:06:59,963 INFO [streaming_decode.py:582] Cuts processed until now is 4900.
155
+ 2025-11-17 17:07:04,160 INFO [streaming_decode.py:582] Cuts processed until now is 5000.
156
+ 2025-11-17 17:07:08,749 INFO [streaming_decode.py:582] Cuts processed until now is 5100.
157
+ 2025-11-17 17:07:12,851 INFO [streaming_decode.py:582] Cuts processed until now is 5200.
158
+ 2025-11-17 17:07:17,577 INFO [streaming_decode.py:582] Cuts processed until now is 5300.
159
+ 2025-11-17 17:07:22,027 INFO [streaming_decode.py:582] Cuts processed until now is 5400.
160
+ 2025-11-17 17:07:26,196 INFO [streaming_decode.py:582] Cuts processed until now is 5500.
161
+ 2025-11-17 17:07:30,520 INFO [streaming_decode.py:582] Cuts processed until now is 5600.
162
+ 2025-11-17 17:07:34,676 INFO [streaming_decode.py:582] Cuts processed until now is 5700.
163
+ 2025-11-17 17:07:38,973 INFO [streaming_decode.py:582] Cuts processed until now is 5800.
164
+ 2025-11-17 17:07:47,012 INFO [streaming_decode.py:582] Cuts processed until now is 5900.
165
+ 2025-11-17 17:07:51,273 INFO [streaming_decode.py:582] Cuts processed until now is 6000.
166
+ 2025-11-17 17:07:55,845 INFO [streaming_decode.py:582] Cuts processed until now is 6100.
167
+ 2025-11-17 17:08:00,199 INFO [streaming_decode.py:582] Cuts processed until now is 6200.
168
+ 2025-11-17 17:08:04,507 INFO [streaming_decode.py:582] Cuts processed until now is 6300.
169
+ 2025-11-17 17:08:08,944 INFO [streaming_decode.py:582] Cuts processed until now is 6400.
170
+ 2025-11-17 17:08:13,132 INFO [streaming_decode.py:582] Cuts processed until now is 6500.
171
+ 2025-11-17 17:08:16,990 INFO [streaming_decode.py:582] Cuts processed until now is 6600.
172
+ 2025-11-17 17:08:21,302 INFO [streaming_decode.py:582] Cuts processed until now is 6700.
173
+ 2025-11-17 17:08:25,809 INFO [streaming_decode.py:582] Cuts processed until now is 6800.
174
+ 2025-11-17 17:08:29,921 INFO [streaming_decode.py:582] Cuts processed until now is 6900.
175
+ 2025-11-17 17:08:33,806 INFO [streaming_decode.py:582] Cuts processed until now is 7000.
176
+ 2025-11-17 17:08:38,312 INFO [streaming_decode.py:582] Cuts processed until now is 7100.
177
+ 2025-11-17 17:08:45,834 INFO [streaming_decode.py:582] Cuts processed until now is 7200.
178
+ 2025-11-17 17:08:50,154 INFO [streaming_decode.py:582] Cuts processed until now is 7300.
179
+ 2025-11-17 17:08:54,556 INFO [streaming_decode.py:582] Cuts processed until now is 7400.
180
+ 2025-11-17 17:08:58,864 INFO [streaming_decode.py:582] Cuts processed until now is 7500.
181
+ 2025-11-17 17:09:03,258 INFO [streaming_decode.py:582] Cuts processed until now is 7600.
182
+ 2025-11-17 17:09:07,734 INFO [streaming_decode.py:582] Cuts processed until now is 7700.
183
+ 2025-11-17 17:09:12,054 INFO [streaming_decode.py:582] Cuts processed until now is 7800.
184
+ 2025-11-17 17:09:16,219 INFO [streaming_decode.py:582] Cuts processed until now is 7900.
185
+ 2025-11-17 17:09:21,208 INFO [streaming_decode.py:582] Cuts processed until now is 8000.
186
+ 2025-11-17 17:09:25,527 INFO [streaming_decode.py:582] Cuts processed until now is 8100.
187
+ 2025-11-17 17:09:30,358 INFO [streaming_decode.py:582] Cuts processed until now is 8200.
188
+ 2025-11-17 17:09:35,343 INFO [streaming_decode.py:582] Cuts processed until now is 8300.
189
+ 2025-11-17 17:09:40,129 INFO [streaming_decode.py:582] Cuts processed until now is 8400.
190
+ 2025-11-17 17:09:44,775 INFO [streaming_decode.py:582] Cuts processed until now is 8500.
191
+ 2025-11-17 17:09:49,104 INFO [streaming_decode.py:582] Cuts processed until now is 8600.
192
+ 2025-11-17 17:09:53,815 INFO [streaming_decode.py:582] Cuts processed until now is 8700.
193
+ 2025-11-17 17:10:01,892 INFO [streaming_decode.py:582] Cuts processed until now is 8800.
194
+ 2025-11-17 17:10:05,935 INFO [streaming_decode.py:582] Cuts processed until now is 8900.
195
+ 2025-11-17 17:10:10,717 INFO [streaming_decode.py:582] Cuts processed until now is 9000.
196
+ 2025-11-17 17:10:15,146 INFO [streaming_decode.py:582] Cuts processed until now is 9100.
197
+ 2025-11-17 17:10:19,535 INFO [streaming_decode.py:582] Cuts processed until now is 9200.
198
+ 2025-11-17 17:10:24,122 INFO [streaming_decode.py:582] Cuts processed until now is 9300.
199
+ 2025-11-17 17:10:28,970 INFO [streaming_decode.py:582] Cuts processed until now is 9400.
200
+ 2025-11-17 17:10:33,520 INFO [streaming_decode.py:582] Cuts processed until now is 9500.
201
+ 2025-11-17 17:10:37,372 INFO [streaming_decode.py:582] Cuts processed until now is 9600.
202
+ 2025-11-17 17:10:41,616 INFO [streaming_decode.py:582] Cuts processed until now is 9700.
203
+ 2025-11-17 17:10:46,001 INFO [streaming_decode.py:582] Cuts processed until now is 9800.
204
+ 2025-11-17 17:10:49,921 INFO [streaming_decode.py:582] Cuts processed until now is 9900.
205
+ 2025-11-17 17:10:53,930 INFO [streaming_decode.py:582] Cuts processed until now is 10000.
206
+ 2025-11-17 17:10:58,105 INFO [streaming_decode.py:582] Cuts processed until now is 10100.
207
+ 2025-11-17 17:11:01,767 INFO [streaming_decode.py:582] Cuts processed until now is 10200.
208
+ 2025-11-17 17:11:09,065 INFO [streaming_decode.py:582] Cuts processed until now is 10300.
209
+ 2025-11-17 17:11:13,222 INFO [streaming_decode.py:582] Cuts processed until now is 10400.
210
+ 2025-11-17 17:11:17,764 INFO [streaming_decode.py:582] Cuts processed until now is 10500.
211
+ 2025-11-17 17:11:22,499 INFO [streaming_decode.py:582] Cuts processed until now is 10600.
212
+ 2025-11-17 17:11:26,685 INFO [streaming_decode.py:582] Cuts processed until now is 10700.
213
+ 2025-11-17 17:11:30,592 INFO [streaming_decode.py:582] Cuts processed until now is 10800.
214
+ 2025-11-17 17:11:35,342 INFO [streaming_decode.py:582] Cuts processed until now is 10900.
215
+ 2025-11-17 17:11:39,864 INFO [streaming_decode.py:582] Cuts processed until now is 11000.
216
+ 2025-11-17 17:11:44,131 INFO [streaming_decode.py:582] Cuts processed until now is 11100.
217
+ 2025-11-17 17:11:48,516 INFO [streaming_decode.py:582] Cuts processed until now is 11200.
218
+ 2025-11-17 17:11:53,327 INFO [streaming_decode.py:582] Cuts processed until now is 11300.
219
+ 2025-11-17 17:11:58,032 INFO [streaming_decode.py:582] Cuts processed until now is 11400.
220
+ 2025-11-17 17:12:05,981 INFO [streaming_decode.py:582] Cuts processed until now is 11500.
221
+ 2025-11-17 17:12:10,686 INFO [streaming_decode.py:582] Cuts processed until now is 11600.
222
+ 2025-11-17 17:12:15,139 INFO [streaming_decode.py:582] Cuts processed until now is 11700.
223
+ 2025-11-17 17:12:19,187 INFO [streaming_decode.py:582] Cuts processed until now is 11800.
224
+ 2025-11-17 17:12:23,610 INFO [streaming_decode.py:582] Cuts processed until now is 11900.
225
+ 2025-11-17 17:12:27,852 INFO [streaming_decode.py:582] Cuts processed until now is 12000.
226
+ 2025-11-17 17:12:32,329 INFO [streaming_decode.py:582] Cuts processed until now is 12100.
227
+ 2025-11-17 17:12:36,476 INFO [streaming_decode.py:582] Cuts processed until now is 12200.
228
+ 2025-11-17 17:12:40,712 INFO [streaming_decode.py:582] Cuts processed until now is 12300.
229
+ 2025-11-17 17:12:44,766 INFO [streaming_decode.py:582] Cuts processed until now is 12400.
230
+ 2025-11-17 17:12:48,957 INFO [streaming_decode.py:582] Cuts processed until now is 12500.
231
+ 2025-11-17 17:12:53,499 INFO [streaming_decode.py:582] Cuts processed until now is 12600.
232
+ 2025-11-17 17:12:57,751 INFO [streaming_decode.py:582] Cuts processed until now is 12700.
233
+ 2025-11-17 17:13:02,017 INFO [streaming_decode.py:582] Cuts processed until now is 12800.
234
+ 2025-11-17 17:13:06,366 INFO [streaming_decode.py:582] Cuts processed until now is 12900.
235
+ 2025-11-17 17:13:10,328 INFO [streaming_decode.py:582] Cuts processed until now is 13000.
236
+ 2025-11-17 17:13:14,027 INFO [streaming_decode.py:582] Cuts processed until now is 13100.
237
+ 2025-11-17 17:13:22,086 INFO [streaming_decode.py:582] Cuts processed until now is 13200.
238
+ 2025-11-17 17:13:26,418 INFO [streaming_decode.py:582] Cuts processed until now is 13300.
239
+ 2025-11-17 17:13:31,102 INFO [streaming_decode.py:582] Cuts processed until now is 13400.
240
+ 2025-11-17 17:13:35,645 INFO [streaming_decode.py:582] Cuts processed until now is 13500.
241
+ 2025-11-17 17:13:39,907 INFO [streaming_decode.py:582] Cuts processed until now is 13600.
242
+ 2025-11-17 17:13:44,914 INFO [streaming_decode.py:582] Cuts processed until now is 13700.
243
+ 2025-11-17 17:13:48,618 INFO [streaming_decode.py:582] Cuts processed until now is 13800.
244
+ 2025-11-17 17:13:53,176 INFO [streaming_decode.py:582] Cuts processed until now is 13900.
245
+ 2025-11-17 17:13:58,002 INFO [streaming_decode.py:582] Cuts processed until now is 14000.
246
+ 2025-11-17 17:14:01,878 INFO [streaming_decode.py:582] Cuts processed until now is 14100.
247
+ 2025-11-17 17:14:06,372 INFO [streaming_decode.py:582] Cuts processed until now is 14200.
248
+ 2025-11-17 17:14:10,416 INFO [streaming_decode.py:582] Cuts processed until now is 14300.
249
+ 2025-11-17 17:14:17,712 INFO [streaming_decode.py:582] Cuts processed until now is 14400.
250
+ 2025-11-17 17:14:21,844 INFO [streaming_decode.py:582] Cuts processed until now is 14500.
251
+ 2025-11-17 17:14:26,011 INFO [streaming_decode.py:582] Cuts processed until now is 14600.
252
+ 2025-11-17 17:14:30,577 INFO [streaming_decode.py:582] Cuts processed until now is 14700.
253
+ 2025-11-17 17:14:34,656 INFO [streaming_decode.py:582] Cuts processed until now is 14800.
254
+ 2025-11-17 17:14:38,905 INFO [streaming_decode.py:582] Cuts processed until now is 14900.
255
+ 2025-11-17 17:14:42,996 INFO [streaming_decode.py:582] Cuts processed until now is 15000.
256
+ 2025-11-17 17:14:47,167 INFO [streaming_decode.py:582] Cuts processed until now is 15100.
257
+ 2025-11-17 17:14:51,035 INFO [streaming_decode.py:582] Cuts processed until now is 15200.
258
+ 2025-11-17 17:14:55,440 INFO [streaming_decode.py:582] Cuts processed until now is 15300.
259
+ 2025-11-17 17:14:59,749 INFO [streaming_decode.py:582] Cuts processed until now is 15400.
260
+ 2025-11-17 17:15:06,889 INFO [streaming_decode.py:582] Cuts processed until now is 15500.
261
+ 2025-11-17 17:15:07,563 INFO [streaming_decode.py:582] Cuts processed until now is 15600.
262
+ 2025-11-17 17:15:11,953 INFO [streaming_decode.py:582] Cuts processed until now is 15700.
263
+ 2025-11-17 17:15:16,524 INFO [streaming_decode.py:582] Cuts processed until now is 15800.
264
+ 2025-11-17 17:15:45,351 INFO [streaming_decode.py:618] The transcripts are stored in tmp/exp-causal-80-epoch/streaming/modified_beam_search/recogs-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt
265
+ 2025-11-17 17:15:45,682 INFO [utils.py:670] [test-commonvoice-num_active_paths_4] %WER 2.45% [19055 / 778666, 3651 ins, 7180 del, 8224 sub ]
266
+ 2025-11-17 17:15:46,506 INFO [streaming_decode.py:627] Wrote detailed error stats to tmp/exp-causal-80-epoch/streaming/modified_beam_search/errs-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt
267
+ 2025-11-17 17:15:46,507 INFO [streaming_decode.py:641]
268
+ For test-commonvoice, WER of different settings are:
269
+ num_active_paths_4 2.45 best for test-commonvoice
270
+
271
+ 2025-11-17 17:15:46,512 INFO [streaming_decode.py:582] Cuts processed until now is 0.
272
+ 2025-11-17 17:15:47,324 INFO [streaming_decode.py:582] Cuts processed until now is 100.
273
+ 2025-11-17 17:15:48,183 INFO [streaming_decode.py:582] Cuts processed until now is 200.
274
+ 2025-11-17 17:15:48,884 INFO [streaming_decode.py:582] Cuts processed until now is 300.
275
+ 2025-11-17 17:15:49,818 INFO [streaming_decode.py:582] Cuts processed until now is 400.
276
+ 2025-11-17 17:15:50,758 INFO [streaming_decode.py:582] Cuts processed until now is 500.
277
+ 2025-11-17 17:15:51,707 INFO [streaming_decode.py:582] Cuts processed until now is 600.
278
+ 2025-11-17 17:15:52,668 INFO [streaming_decode.py:582] Cuts processed until now is 700.
279
+ 2025-11-17 17:15:53,535 INFO [streaming_decode.py:582] Cuts processed until now is 800.
280
+ 2025-11-17 17:15:54,373 INFO [streaming_decode.py:582] Cuts processed until now is 900.
281
+ 2025-11-17 17:16:29,747 INFO [streaming_decode.py:618] The transcripts are stored in tmp/exp-causal-80-epoch/streaming/modified_beam_search/recogs-test-slr72-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt
282
+ 2025-11-17 17:16:29,765 INFO [utils.py:670] [test-slr72-num_active_paths_4] %WER 1.12% [456 / 40600, 75 ins, 240 del, 141 sub ]
283
+ 2025-11-17 17:16:29,806 INFO [streaming_decode.py:627] Wrote detailed error stats to tmp/exp-causal-80-epoch/streaming/modified_beam_search/errs-test-slr72-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt
284
+ 2025-11-17 17:16:29,807 INFO [streaming_decode.py:641]
285
+ For test-slr72, WER of different settings are:
286
+ num_active_paths_4 1.12 best for test-slr72
287
+
288
+ 2025-11-17 17:16:29,807 INFO [streaming_decode.py:794] Done!
streaming/modified_beam_search/recogs-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/modified_beam_search/recogs-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/modified_beam_search/recogs-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/modified_beam_search/recogs-test-slr72-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt ADDED
The diff for this file is too large to render. See raw diff
 
streaming/modified_beam_search/wer-summary-test-commonvoice-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ settings WER
2
+ num_active_paths_4 2.71
streaming/modified_beam_search/wer-summary-test-commonvoice-epoch-80-avg-5-chunk-32-left-context-128-use-averaged-model.txt ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ settings WER
2
+ num_active_paths_4 2.45
streaming/modified_beam_search/wer-summary-test-slr72-epoch-80-avg-5-chunk-16-left-context-128-use-averaged-model.txt ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ settings WER
2
+ num_active_paths_4 1.47