Hugging Face's logo

speechcolab
/
icefall-asr-gigaspeech-conformer-ctc

Automatic Speech Recognition
K2
English
icefall
audio
Model card Files
xet
Community
wgb14 commited on
Commit
81b7e35
·
1 Parent(s): e08bf6b

Upload log/log-decode-2022-04-09-01-40-41

Browse files
Files changed (1) hide show
  1. log/log-decode-2022-04-09-01-40-41 +1176 -0
log/log-decode-2022-04-09-01-40-41 ADDED
@@ -0,0 +1,1176 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 2022-04-09 01:40:41,909 INFO [decode_test.py:583] Decoding started
2
+ 2022-04-09 01:40:41,910 INFO [decode_test.py:584] {'subsampling_factor': 4, 'vgg_frontend': False, 'use_feat_batchnorm': True, 'feature_dim': 80, 'nhead': 8, 'attention_dim': 512, 'num_decoder_layers': 6, 'search_beam': 20, 'output_beam': 8, 'min_active_states': 30, 'max_active_states': 10000, 'use_double_scores': True, 'env_info': {'k2-version': '1.14', 'k2-build-type': 'Release', 'k2-with-cuda': True, 'k2-git-sha1': '6833270cb228aba7bf9681fccd41e2b52f7d984c', 'k2-git-date': 'Wed Mar 16 11:16:05 2022', 'lhotse-version': '1.0.0.dev+git.d917411.clean', 'torch-cuda-available': True, 'torch-cuda-version': '11.1', 'python-version': '3.7', 'icefall-git-branch': 'gigaspeech_recipe', 'icefall-git-sha1': 'c3993a5-dirty', 'icefall-git-date': 'Mon Mar 21 13:49:39 2022', 'icefall-path': '/userhome/user/guanbo/icefall_decode', 'k2-path': '/opt/conda/lib/python3.7/site-packages/k2-1.14.dev20220408+cuda11.1.torch1.10.0-py3.7-linux-x86_64.egg/k2/__init__.py', 'lhotse-path': '/userhome/user/guanbo/lhotse/lhotse/__init__.py', 'hostname': 'c8861f400b70d011ec0a3ee069db84328338-chenx8564-0', 'IP address': '10.9.150.55'}, 'epoch': 18, 'avg': 6, 'method': 'attention-decoder', 'num_paths': 1000, 'nbest_scale': 0.5, 'exp_dir': PosixPath('conformer_ctc/exp_500_8_2'), 'lang_dir': PosixPath('data/lang_bpe_500'), 'lm_dir': PosixPath('data/lm'), 'manifest_dir': PosixPath('data/fbank'), 'max_duration': 20, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'on_the_fly_feats': False, 'shuffle': True, 'return_cuts': True, 'num_workers': 1, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'subset': 'XL', 'lazy_load': True, 'small_dev': False}
3
+ 2022-04-09 01:40:42,371 INFO [lexicon.py:176] Loading pre-compiled data/lang_bpe_500/Linv.pt
4
+ 2022-04-09 01:40:42,473 INFO [decode_test.py:594] device: cuda:0
5
+ 2022-04-09 01:40:46,249 INFO [decode_test.py:656] Loading pre-compiled G_4_gram.pt
6
+ 2022-04-09 01:40:47,406 INFO [decode_test.py:692] averaging ['conformer_ctc/exp_500_8_2/epoch-13.pt', 'conformer_ctc/exp_500_8_2/epoch-14.pt', 'conformer_ctc/exp_500_8_2/epoch-15.pt', 'conformer_ctc/exp_500_8_2/epoch-16.pt', 'conformer_ctc/exp_500_8_2/epoch-17.pt', 'conformer_ctc/exp_500_8_2/epoch-18.pt']
7
+ 2022-04-09 01:40:53,065 INFO [decode_test.py:699] Number of model parameters: 109226120
8
+ 2022-04-09 01:40:53,065 INFO [asr_datamodule.py:381] About to get test cuts
9
+ 2022-04-09 01:40:56,361 INFO [decode_test.py:497] batch 0/?, cuts processed until now is 3
10
+ 2022-04-09 01:41:24,462 INFO [decode.py:736] Caught exception:
11
+ CUDA out of memory. Tried to allocate 5.93 GiB (GPU 0; 31.75 GiB total capacity; 27.23 GiB already allocated; 1.90 GiB free; 28.49 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
12
+
13
+ 2022-04-09 01:41:24,462 INFO [decode.py:743] num_arcs before pruning: 324363
14
+ 2022-04-09 01:41:24,462 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
15
+ 2022-04-09 01:41:24,473 INFO [decode.py:757] num_arcs after pruning: 7174
16
+ 2022-04-09 01:41:40,284 INFO [decode.py:736] Caught exception:
17
+ CUDA out of memory. Tried to allocate 4.67 GiB (GPU 0; 31.75 GiB total capacity; 25.69 GiB already allocated; 2.92 GiB free; 27.47 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
18
+
19
+ 2022-04-09 01:41:40,285 INFO [decode.py:743] num_arcs before pruning: 368362
20
+ 2022-04-09 01:41:40,285 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
21
+ 2022-04-09 01:41:40,305 INFO [decode.py:757] num_arcs after pruning: 8521
22
+ 2022-04-09 01:42:38,727 INFO [decode.py:736] Caught exception:
23
+ CUDA out of memory. Tried to allocate 2.18 GiB (GPU 0; 31.75 GiB total capacity; 26.05 GiB already allocated; 1.42 GiB free; 28.98 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
24
+
25
+ 2022-04-09 01:42:38,727 INFO [decode.py:743] num_arcs before pruning: 432616
26
+ 2022-04-09 01:42:38,728 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
27
+ 2022-04-09 01:42:38,736 INFO [decode.py:757] num_arcs after pruning: 9233
28
+ 2022-04-09 01:43:13,573 INFO [decode_test.py:497] batch 100/?, cuts processed until now is 297
29
+ 2022-04-09 01:43:48,362 INFO [decode.py:736] Caught exception:
30
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 25.34 GiB already allocated; 2.20 GiB free; 28.20 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
31
+
32
+ 2022-04-09 01:43:48,363 INFO [decode.py:743] num_arcs before pruning: 319907
33
+ 2022-04-09 01:43:48,363 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
34
+ 2022-04-09 01:43:48,372 INFO [decode.py:757] num_arcs after pruning: 6358
35
+ 2022-04-09 01:43:59,713 INFO [decode.py:736] Caught exception:
36
+ CUDA out of memory. Tried to allocate 2.74 GiB (GPU 0; 31.75 GiB total capacity; 27.51 GiB already allocated; 2.19 GiB free; 28.20 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
37
+
38
+ 2022-04-09 01:43:59,713 INFO [decode.py:743] num_arcs before pruning: 313596
39
+ 2022-04-09 01:43:59,713 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
40
+ 2022-04-09 01:43:59,724 INFO [decode.py:757] num_arcs after pruning: 8252
41
+ 2022-04-09 01:44:54,463 INFO [decode.py:736] Caught exception:
42
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 25.25 GiB already allocated; 3.07 GiB free; 27.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
43
+
44
+ 2022-04-09 01:44:54,463 INFO [decode.py:743] num_arcs before pruning: 353355
45
+ 2022-04-09 01:44:54,463 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
46
+ 2022-04-09 01:44:54,485 INFO [decode.py:757] num_arcs after pruning: 7520
47
+ 2022-04-09 01:45:20,716 INFO [decode_test.py:497] batch 200/?, cuts processed until now is 570
48
+ 2022-04-09 01:47:19,457 INFO [decode_test.py:497] batch 300/?, cuts processed until now is 806
49
+ 2022-04-09 01:47:38,292 INFO [decode.py:736] Caught exception:
50
+ CUDA out of memory. Tried to allocate 2.28 GiB (GPU 0; 31.75 GiB total capacity; 26.28 GiB already allocated; 1.48 GiB free; 28.92 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
51
+
52
+ 2022-04-09 01:47:38,293 INFO [decode.py:743] num_arcs before pruning: 596002
53
+ 2022-04-09 01:47:38,293 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
54
+ 2022-04-09 01:47:38,312 INFO [decode.py:757] num_arcs after pruning: 10745
55
+ 2022-04-09 01:49:18,493 INFO [decode.py:736] Caught exception:
56
+
57
+ Some bad things happened. Please read the above error messages and stack
58
+ trace. If you are using Python, the following command may be helpful:
59
+
60
+ gdb --args python /path/to/your/code.py
61
+
62
+ (You can use `gdb` to debug the code. Please consider compiling
63
+ a debug version of k2.).
64
+
65
+ If you are unable to fix it, please open an issue at:
66
+
67
+ https://github.com/k2-fsa/k2/issues/new
68
+
69
+
70
+ 2022-04-09 01:49:18,494 INFO [decode.py:743] num_arcs before pruning: 398202
71
+ 2022-04-09 01:49:18,494 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
72
+ 2022-04-09 01:49:18,541 INFO [decode.py:757] num_arcs after pruning: 14003
73
+ 2022-04-09 01:49:21,800 INFO [decode_test.py:497] batch 400/?, cuts processed until now is 1082
74
+ 2022-04-09 01:50:58,700 INFO [decode.py:736] Caught exception:
75
+ CUDA out of memory. Tried to allocate 4.85 GiB (GPU 0; 31.75 GiB total capacity; 25.89 GiB already allocated; 1.48 GiB free; 28.92 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
76
+
77
+ 2022-04-09 01:50:58,701 INFO [decode.py:743] num_arcs before pruning: 398349
78
+ 2022-04-09 01:50:58,701 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
79
+ 2022-04-09 01:50:58,709 INFO [decode.py:757] num_arcs after pruning: 10321
80
+ 2022-04-09 01:51:31,627 INFO [decode_test.py:497] batch 500/?, cuts processed until now is 1334
81
+ 2022-04-09 01:52:05,232 INFO [decode.py:736] Caught exception:
82
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.62 GiB already allocated; 1.47 GiB free; 28.93 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
83
+
84
+ 2022-04-09 01:52:05,232 INFO [decode.py:743] num_arcs before pruning: 212665
85
+ 2022-04-09 01:52:05,232 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
86
+ 2022-04-09 01:52:05,241 INFO [decode.py:757] num_arcs after pruning: 6301
87
+ 2022-04-09 01:53:29,890 INFO [decode.py:736] Caught exception:
88
+ CUDA out of memory. Tried to allocate 1.91 GiB (GPU 0; 31.75 GiB total capacity; 25.66 GiB already allocated; 1.48 GiB free; 28.92 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
89
+
90
+ 2022-04-09 01:53:29,891 INFO [decode.py:743] num_arcs before pruning: 883555
91
+ 2022-04-09 01:53:29,891 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
92
+ 2022-04-09 01:53:29,905 INFO [decode.py:757] num_arcs after pruning: 14819
93
+ 2022-04-09 01:53:38,676 INFO [decode_test.py:497] batch 600/?, cuts processed until now is 1651
94
+ 2022-04-09 01:54:57,438 INFO [decode.py:736] Caught exception:
95
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 25.34 GiB already allocated; 1.48 GiB free; 28.92 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
96
+
97
+ 2022-04-09 01:54:57,438 INFO [decode.py:743] num_arcs before pruning: 515795
98
+ 2022-04-09 01:54:57,438 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
99
+ 2022-04-09 01:54:57,447 INFO [decode.py:757] num_arcs after pruning: 10132
100
+ 2022-04-09 01:55:28,356 INFO [decode.py:736] Caught exception:
101
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.46 GiB already allocated; 1.48 GiB free; 28.92 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
102
+
103
+ 2022-04-09 01:55:28,356 INFO [decode.py:743] num_arcs before pruning: 670748
104
+ 2022-04-09 01:55:28,356 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
105
+ 2022-04-09 01:55:28,365 INFO [decode.py:757] num_arcs after pruning: 10497
106
+ 2022-04-09 01:55:42,238 INFO [decode_test.py:497] batch 700/?, cuts processed until now is 1956
107
+ 2022-04-09 01:57:57,456 INFO [decode_test.py:497] batch 800/?, cuts processed until now is 2238
108
+ 2022-04-09 01:58:04,281 INFO [decode.py:736] Caught exception:
109
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.45 GiB already allocated; 3.07 GiB free; 27.33 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
110
+
111
+ 2022-04-09 01:58:04,282 INFO [decode.py:743] num_arcs before pruning: 175423
112
+ 2022-04-09 01:58:04,282 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
113
+ 2022-04-09 01:58:04,296 INFO [decode.py:757] num_arcs after pruning: 7926
114
+ 2022-04-09 01:59:07,916 INFO [decode.py:736] Caught exception:
115
+ CUDA out of memory. Tried to allocate 4.68 GiB (GPU 0; 31.75 GiB total capacity; 24.40 GiB already allocated; 3.06 GiB free; 27.33 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
116
+
117
+ 2022-04-09 01:59:07,917 INFO [decode.py:743] num_arcs before pruning: 259758
118
+ 2022-04-09 01:59:07,917 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
119
+ 2022-04-09 01:59:07,928 INFO [decode.py:757] num_arcs after pruning: 6026
120
+ 2022-04-09 02:00:00,623 INFO [decode_test.py:497] batch 900/?, cuts processed until now is 2536
121
+ 2022-04-09 02:01:22,959 INFO [decode.py:736] Caught exception:
122
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.44 GiB already allocated; 3.08 GiB free; 27.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
123
+
124
+ 2022-04-09 02:01:22,959 INFO [decode.py:743] num_arcs before pruning: 749228
125
+ 2022-04-09 02:01:22,959 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
126
+ 2022-04-09 02:01:22,968 INFO [decode.py:757] num_arcs after pruning: 23868
127
+ 2022-04-09 02:01:59,449 INFO [decode_test.py:497] batch 1000/?, cuts processed until now is 2824
128
+ 2022-04-09 02:03:05,494 INFO [decode.py:736] Caught exception:
129
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.38 GiB already allocated; 3.06 GiB free; 27.33 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
130
+
131
+ 2022-04-09 02:03:05,494 INFO [decode.py:743] num_arcs before pruning: 255135
132
+ 2022-04-09 02:03:05,494 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
133
+ 2022-04-09 02:03:05,504 INFO [decode.py:757] num_arcs after pruning: 5955
134
+ 2022-04-09 02:03:48,017 INFO [decode.py:736] Caught exception:
135
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.61 GiB already allocated; 3.08 GiB free; 27.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
136
+
137
+ 2022-04-09 02:03:48,017 INFO [decode.py:743] num_arcs before pruning: 517077
138
+ 2022-04-09 02:03:48,017 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
139
+ 2022-04-09 02:03:48,026 INFO [decode.py:757] num_arcs after pruning: 7695
140
+ 2022-04-09 02:04:09,806 INFO [decode_test.py:497] batch 1100/?, cuts processed until now is 3105
141
+ 2022-04-09 02:04:31,410 INFO [decode.py:736] Caught exception:
142
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.34 GiB already allocated; 3.08 GiB free; 27.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
143
+
144
+ 2022-04-09 02:04:31,411 INFO [decode.py:743] num_arcs before pruning: 859561
145
+ 2022-04-09 02:04:31,411 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
146
+ 2022-04-09 02:04:31,422 INFO [decode.py:757] num_arcs after pruning: 13014
147
+ 2022-04-09 02:06:11,496 INFO [decode_test.py:497] batch 1200/?, cuts processed until now is 3401
148
+ 2022-04-09 02:08:10,727 INFO [decode_test.py:497] batch 1300/?, cuts processed until now is 3730
149
+ 2022-04-09 02:10:17,677 INFO [decode_test.py:497] batch 1400/?, cuts processed until now is 4067
150
+ 2022-04-09 02:12:13,175 INFO [decode_test.py:497] batch 1500/?, cuts processed until now is 4329
151
+ 2022-04-09 02:13:02,842 INFO [decode.py:736] Caught exception:
152
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.55 GiB already allocated; 3.08 GiB free; 27.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
153
+
154
+ 2022-04-09 02:13:02,843 INFO [decode.py:743] num_arcs before pruning: 475511
155
+ 2022-04-09 02:13:02,843 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
156
+ 2022-04-09 02:13:02,849 INFO [decode.py:757] num_arcs after pruning: 8439
157
+ 2022-04-09 02:13:46,588 INFO [decode.py:736] Caught exception:
158
+ CUDA out of memory. Tried to allocate 2.37 GiB (GPU 0; 31.75 GiB total capacity; 26.83 GiB already allocated; 1.45 GiB free; 28.94 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
159
+
160
+ 2022-04-09 02:13:46,588 INFO [decode.py:743] num_arcs before pruning: 595488
161
+ 2022-04-09 02:13:46,588 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
162
+ 2022-04-09 02:13:46,598 INFO [decode.py:757] num_arcs after pruning: 13475
163
+ 2022-04-09 02:14:21,206 INFO [decode_test.py:497] batch 1600/?, cuts processed until now is 4598
164
+ 2022-04-09 02:16:42,740 INFO [decode_test.py:497] batch 1700/?, cuts processed until now is 4969
165
+ 2022-04-09 02:17:13,672 INFO [decode.py:736] Caught exception:
166
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 25.39 GiB already allocated; 1.45 GiB free; 28.94 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
167
+
168
+ 2022-04-09 02:17:13,673 INFO [decode.py:743] num_arcs before pruning: 615734
169
+ 2022-04-09 02:17:13,673 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
170
+ 2022-04-09 02:17:13,685 INFO [decode.py:757] num_arcs after pruning: 8684
171
+ 2022-04-09 02:18:54,514 INFO [decode_test.py:497] batch 1800/?, cuts processed until now is 5260
172
+ 2022-04-09 02:18:59,938 INFO [decode.py:736] Caught exception:
173
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.36 GiB already allocated; 3.08 GiB free; 27.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
174
+
175
+ 2022-04-09 02:18:59,938 INFO [decode.py:743] num_arcs before pruning: 360099
176
+ 2022-04-09 02:18:59,938 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
177
+ 2022-04-09 02:18:59,949 INFO [decode.py:757] num_arcs after pruning: 6898
178
+ 2022-04-09 02:19:48,186 INFO [decode.py:736] Caught exception:
179
+ CUDA out of memory. Tried to allocate 6.00 GiB (GPU 0; 31.75 GiB total capacity; 27.15 GiB already allocated; 967.75 MiB free; 29.45 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
180
+
181
+ 2022-04-09 02:19:48,186 INFO [decode.py:743] num_arcs before pruning: 168720
182
+ 2022-04-09 02:19:48,186 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
183
+ 2022-04-09 02:19:48,201 INFO [decode.py:757] num_arcs after pruning: 5346
184
+ 2022-04-09 02:20:52,049 INFO [decode_test.py:497] batch 1900/?, cuts processed until now is 5585
185
+ 2022-04-09 02:22:12,107 INFO [decode.py:736] Caught exception:
186
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.45 GiB already allocated; 973.75 MiB free; 29.44 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
187
+
188
+ 2022-04-09 02:22:12,107 INFO [decode.py:743] num_arcs before pruning: 1151735
189
+ 2022-04-09 02:22:12,107 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
190
+ 2022-04-09 02:22:12,120 INFO [decode.py:757] num_arcs after pruning: 8335
191
+ 2022-04-09 02:23:01,497 INFO [decode_test.py:497] batch 2000/?, cuts processed until now is 5902
192
+ 2022-04-09 02:25:26,356 INFO [decode_test.py:497] batch 2100/?, cuts processed until now is 6219
193
+ 2022-04-09 02:25:56,466 INFO [decode.py:736] Caught exception:
194
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.34 GiB already allocated; 973.75 MiB free; 29.44 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
195
+
196
+ 2022-04-09 02:25:56,467 INFO [decode.py:743] num_arcs before pruning: 612804
197
+ 2022-04-09 02:25:56,467 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
198
+ 2022-04-09 02:25:56,477 INFO [decode.py:757] num_arcs after pruning: 10853
199
+ 2022-04-09 02:27:26,441 INFO [decode_test.py:497] batch 2200/?, cuts processed until now is 6480
200
+ 2022-04-09 02:29:28,073 INFO [decode_test.py:497] batch 2300/?, cuts processed until now is 6768
201
+ 2022-04-09 02:31:41,553 INFO [decode_test.py:497] batch 2400/?, cuts processed until now is 7120
202
+ 2022-04-09 02:31:55,632 INFO [decode.py:736] Caught exception:
203
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.42 GiB already allocated; 3.07 GiB free; 27.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
204
+
205
+ 2022-04-09 02:31:55,632 INFO [decode.py:743] num_arcs before pruning: 411490
206
+ 2022-04-09 02:31:55,632 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
207
+ 2022-04-09 02:31:55,638 INFO [decode.py:757] num_arcs after pruning: 8626
208
+ 2022-04-09 02:33:22,034 INFO [decode.py:736] Caught exception:
209
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.42 GiB already allocated; 3.07 GiB free; 27.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
210
+
211
+ 2022-04-09 02:33:22,034 INFO [decode.py:743] num_arcs before pruning: 625728
212
+ 2022-04-09 02:33:22,035 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
213
+ 2022-04-09 02:33:22,043 INFO [decode.py:757] num_arcs after pruning: 9502
214
+ 2022-04-09 02:33:37,663 INFO [decode_test.py:497] batch 2500/?, cuts processed until now is 7387
215
+ 2022-04-09 02:34:18,300 INFO [decode.py:736] Caught exception:
216
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.51 GiB already allocated; 3.07 GiB free; 27.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
217
+
218
+ 2022-04-09 02:34:18,301 INFO [decode.py:743] num_arcs before pruning: 1015956
219
+ 2022-04-09 02:34:18,301 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
220
+ 2022-04-09 02:34:18,314 INFO [decode.py:757] num_arcs after pruning: 14404
221
+ 2022-04-09 02:34:20,220 INFO [decode.py:841] Caught exception:
222
+ CUDA out of memory. Tried to allocate 5.58 GiB (GPU 0; 31.75 GiB total capacity; 24.87 GiB already allocated; 3.07 GiB free; 27.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
223
+
224
+ 2022-04-09 02:34:20,221 INFO [decode.py:843] num_paths before decreasing: 1000
225
+ 2022-04-09 02:34:20,221 INFO [decode.py:852] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
226
+ 2022-04-09 02:34:20,221 INFO [decode.py:858] num_paths after decreasing: 500
227
+ 2022-04-09 02:34:40,089 INFO [decode.py:736] Caught exception:
228
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.38 GiB already allocated; 3.07 GiB free; 27.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
229
+
230
+ 2022-04-09 02:34:40,089 INFO [decode.py:743] num_arcs before pruning: 570686
231
+ 2022-04-09 02:34:40,089 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
232
+ 2022-04-09 02:34:40,098 INFO [decode.py:757] num_arcs after pruning: 9182
233
+ 2022-04-09 02:35:50,624 INFO [decode_test.py:497] batch 2600/?, cuts processed until now is 7764
234
+ 2022-04-09 02:36:44,519 INFO [decode.py:736] Caught exception:
235
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.61 GiB already allocated; 3.08 GiB free; 27.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
236
+
237
+ 2022-04-09 02:36:44,519 INFO [decode.py:743] num_arcs before pruning: 1066267
238
+ 2022-04-09 02:36:44,519 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
239
+ 2022-04-09 02:36:44,530 INFO [decode.py:757] num_arcs after pruning: 6963
240
+ 2022-04-09 02:38:18,717 INFO [decode_test.py:497] batch 2700/?, cuts processed until now is 8078
241
+ 2022-04-09 02:40:07,021 INFO [decode.py:736] Caught exception:
242
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.42 GiB already allocated; 3.07 GiB free; 27.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
243
+
244
+ 2022-04-09 02:40:07,022 INFO [decode.py:743] num_arcs before pruning: 1023667
245
+ 2022-04-09 02:40:07,022 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
246
+ 2022-04-09 02:40:07,034 INFO [decode.py:757] num_arcs after pruning: 13090
247
+ 2022-04-09 02:40:25,184 INFO [decode_test.py:497] batch 2800/?, cuts processed until now is 8444
248
+ 2022-04-09 02:41:27,080 INFO [decode.py:736] Caught exception:
249
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.32 GiB already allocated; 3.08 GiB free; 27.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
250
+
251
+ 2022-04-09 02:41:27,080 INFO [decode.py:743] num_arcs before pruning: 739744
252
+ 2022-04-09 02:41:27,080 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
253
+ 2022-04-09 02:41:27,093 INFO [decode.py:757] num_arcs after pruning: 9791
254
+ 2022-04-09 02:42:44,319 INFO [decode_test.py:497] batch 2900/?, cuts processed until now is 8765
255
+ 2022-04-09 02:42:44,656 INFO [decode.py:736] Caught exception:
256
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.73 GiB already allocated; 3.08 GiB free; 27.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
257
+
258
+ 2022-04-09 02:42:44,656 INFO [decode.py:743] num_arcs before pruning: 666168
259
+ 2022-04-09 02:42:44,656 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
260
+ 2022-04-09 02:42:44,665 INFO [decode.py:757] num_arcs after pruning: 17223
261
+ 2022-04-09 02:43:05,748 INFO [decode.py:736] Caught exception:
262
+ CUDA out of memory. Tried to allocate 5.60 GiB (GPU 0; 31.75 GiB total capacity; 26.18 GiB already allocated; 1.14 GiB free; 29.26 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
263
+
264
+ 2022-04-09 02:43:05,748 INFO [decode.py:743] num_arcs before pruning: 188729
265
+ 2022-04-09 02:43:05,748 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
266
+ 2022-04-09 02:43:05,762 INFO [decode.py:757] num_arcs after pruning: 8688
267
+ 2022-04-09 02:44:54,469 INFO [decode_test.py:497] batch 3000/?, cuts processed until now is 9050
268
+ 2022-04-09 02:46:55,167 INFO [decode_test.py:497] batch 3100/?, cuts processed until now is 9296
269
+ 2022-04-09 02:47:28,418 INFO [decode.py:736] Caught exception:
270
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 20.00 GiB already allocated; 3.07 GiB free; 27.33 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
271
+
272
+ 2022-04-09 02:47:28,419 INFO [decode.py:743] num_arcs before pruning: 160153
273
+ 2022-04-09 02:47:28,419 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
274
+ 2022-04-09 02:47:28,448 INFO [decode.py:757] num_arcs after pruning: 7778
275
+ 2022-04-09 02:49:21,448 INFO [decode_test.py:497] batch 3200/?, cuts processed until now is 9652
276
+ 2022-04-09 02:50:17,558 INFO [decode.py:736] Caught exception:
277
+ CUDA out of memory. Tried to allocate 6.13 GiB (GPU 0; 31.75 GiB total capacity; 27.60 GiB already allocated; 895.75 MiB free; 29.52 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
278
+
279
+ 2022-04-09 02:50:17,558 INFO [decode.py:743] num_arcs before pruning: 388116
280
+ 2022-04-09 02:50:17,559 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
281
+ 2022-04-09 02:50:17,565 INFO [decode.py:757] num_arcs after pruning: 10555
282
+ 2022-04-09 02:51:30,675 INFO [decode_test.py:497] batch 3300/?, cuts processed until now is 10071
283
+ 2022-04-09 02:53:49,565 INFO [decode_test.py:497] batch 3400/?, cuts processed until now is 10342
284
+ 2022-04-09 02:55:49,392 INFO [decode_test.py:497] batch 3500/?, cuts processed until now is 10642
285
+ 2022-04-09 02:58:07,518 INFO [decode_test.py:497] batch 3600/?, cuts processed until now is 10951
286
+ 2022-04-09 02:58:16,360 INFO [decode.py:736] Caught exception:
287
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.29 GiB already allocated; 3.07 GiB free; 27.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
288
+
289
+ 2022-04-09 02:58:16,361 INFO [decode.py:743] num_arcs before pruning: 396714
290
+ 2022-04-09 02:58:16,361 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
291
+ 2022-04-09 02:58:16,374 INFO [decode.py:757] num_arcs after pruning: 9543
292
+ 2022-04-09 03:00:00,485 INFO [decode_test.py:497] batch 3700/?, cuts processed until now is 11231
293
+ 2022-04-09 03:00:17,600 INFO [decode.py:736] Caught exception:
294
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.45 GiB already allocated; 3.07 GiB free; 27.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
295
+
296
+ 2022-04-09 03:00:17,601 INFO [decode.py:743] num_arcs before pruning: 854366
297
+ 2022-04-09 03:00:17,601 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
298
+ 2022-04-09 03:00:17,612 INFO [decode.py:757] num_arcs after pruning: 10487
299
+ 2022-04-09 03:00:20,098 INFO [decode.py:736] Caught exception:
300
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.68 GiB already allocated; 3.07 GiB free; 27.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
301
+
302
+ 2022-04-09 03:00:20,098 INFO [decode.py:743] num_arcs before pruning: 442824
303
+ 2022-04-09 03:00:20,098 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
304
+ 2022-04-09 03:00:20,108 INFO [decode.py:757] num_arcs after pruning: 5265
305
+ 2022-04-09 03:02:00,114 INFO [decode_test.py:497] batch 3800/?, cuts processed until now is 11509
306
+ 2022-04-09 03:02:11,570 INFO [decode.py:736] Caught exception:
307
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.19 GiB already allocated; 3.07 GiB free; 27.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
308
+
309
+ 2022-04-09 03:02:11,571 INFO [decode.py:743] num_arcs before pruning: 285638
310
+ 2022-04-09 03:02:11,571 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
311
+ 2022-04-09 03:02:11,579 INFO [decode.py:757] num_arcs after pruning: 5903
312
+ 2022-04-09 03:04:02,757 INFO [decode_test.py:497] batch 3900/?, cuts processed until now is 11774
313
+ 2022-04-09 03:05:19,989 INFO [decode.py:736] Caught exception:
314
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.73 GiB already allocated; 3.08 GiB free; 27.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
315
+
316
+ 2022-04-09 03:05:19,990 INFO [decode.py:743] num_arcs before pruning: 637327
317
+ 2022-04-09 03:05:19,990 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
318
+ 2022-04-09 03:05:19,999 INFO [decode.py:757] num_arcs after pruning: 6357
319
+ 2022-04-09 03:06:01,953 INFO [decode_test.py:497] batch 4000/?, cuts processed until now is 12045
320
+ 2022-04-09 03:07:49,854 INFO [decode_test.py:497] batch 4100/?, cuts processed until now is 12300
321
+ 2022-04-09 03:09:15,137 INFO [decode.py:736] Caught exception:
322
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.45 GiB already allocated; 3.08 GiB free; 27.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
323
+
324
+ 2022-04-09 03:09:15,138 INFO [decode.py:743] num_arcs before pruning: 507733
325
+ 2022-04-09 03:09:15,138 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
326
+ 2022-04-09 03:09:15,148 INFO [decode.py:757] num_arcs after pruning: 4196
327
+ 2022-04-09 03:09:47,397 INFO [decode.py:736] Caught exception:
328
+ CUDA out of memory. Tried to allocate 5.86 GiB (GPU 0; 31.75 GiB total capacity; 27.78 GiB already allocated; 925.75 MiB free; 29.49 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
329
+
330
+ 2022-04-09 03:09:47,397 INFO [decode.py:743] num_arcs before pruning: 514118
331
+ 2022-04-09 03:09:47,397 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
332
+ 2022-04-09 03:09:47,407 INFO [decode.py:757] num_arcs after pruning: 7168
333
+ 2022-04-09 03:10:00,013 INFO [decode_test.py:497] batch 4200/?, cuts processed until now is 12580
334
+ 2022-04-09 03:10:33,411 INFO [decode.py:736] Caught exception:
335
+ CUDA out of memory. Tried to allocate 2.80 GiB (GPU 0; 31.75 GiB total capacity; 27.70 GiB already allocated; 925.75 MiB free; 29.49 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
336
+
337
+ 2022-04-09 03:10:33,411 INFO [decode.py:743] num_arcs before pruning: 374935
338
+ 2022-04-09 03:10:33,411 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
339
+ 2022-04-09 03:10:33,418 INFO [decode.py:757] num_arcs after pruning: 10023
340
+ 2022-04-09 03:12:04,333 INFO [decode_test.py:497] batch 4300/?, cuts processed until now is 12807
341
+ 2022-04-09 03:14:06,889 INFO [decode_test.py:497] batch 4400/?, cuts processed until now is 13050
342
+ 2022-04-09 03:14:34,787 INFO [decode.py:736] Caught exception:
343
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.47 GiB already allocated; 925.75 MiB free; 29.49 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
344
+
345
+ 2022-04-09 03:14:34,788 INFO [decode.py:743] num_arcs before pruning: 767465
346
+ 2022-04-09 03:14:34,788 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
347
+ 2022-04-09 03:14:34,797 INFO [decode.py:757] num_arcs after pruning: 19151
348
+ 2022-04-09 03:15:08,864 INFO [decode.py:736] Caught exception:
349
+
350
+ Some bad things happened. Please read the above error messages and stack
351
+ trace. If you are using Python, the following command may be helpful:
352
+
353
+ gdb --args python /path/to/your/code.py
354
+
355
+ (You can use `gdb` to debug the code. Please consider compiling
356
+ a debug version of k2.).
357
+
358
+ If you are unable to fix it, please open an issue at:
359
+
360
+ https://github.com/k2-fsa/k2/issues/new
361
+
362
+
363
+ 2022-04-09 03:15:08,864 INFO [decode.py:743] num_arcs before pruning: 123833
364
+ 2022-04-09 03:15:08,864 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
365
+ 2022-04-09 03:15:08,913 INFO [decode.py:757] num_arcs after pruning: 4150
366
+ 2022-04-09 03:15:34,899 INFO [decode.py:736] Caught exception:
367
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 25.64 GiB already allocated; 3.07 GiB free; 27.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
368
+
369
+ 2022-04-09 03:15:34,899 INFO [decode.py:743] num_arcs before pruning: 444800
370
+ 2022-04-09 03:15:34,899 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
371
+ 2022-04-09 03:15:34,908 INFO [decode.py:757] num_arcs after pruning: 11839
372
+ 2022-04-09 03:16:08,462 INFO [decode_test.py:497] batch 4500/?, cuts processed until now is 13295
373
+ 2022-04-09 03:17:56,946 INFO [decode_test.py:497] batch 4600/?, cuts processed until now is 13593
374
+ 2022-04-09 03:18:16,099 INFO [decode.py:736] Caught exception:
375
+ CUDA out of memory. Tried to allocate 5.53 GiB (GPU 0; 31.75 GiB total capacity; 26.53 GiB already allocated; 1.12 GiB free; 29.28 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
376
+
377
+ 2022-04-09 03:18:16,099 INFO [decode.py:743] num_arcs before pruning: 350609
378
+ 2022-04-09 03:18:16,100 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
379
+ 2022-04-09 03:18:16,105 INFO [decode.py:757] num_arcs after pruning: 9262
380
+ 2022-04-09 03:19:57,230 INFO [decode_test.py:497] batch 4700/?, cuts processed until now is 13858
381
+ 2022-04-09 03:20:19,775 INFO [decode.py:736] Caught exception:
382
+ CUDA out of memory. Tried to allocate 4.87 GiB (GPU 0; 31.75 GiB total capacity; 25.78 GiB already allocated; 1.12 GiB free; 29.28 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
383
+
384
+ 2022-04-09 03:20:19,775 INFO [decode.py:743] num_arcs before pruning: 375071
385
+ 2022-04-09 03:20:19,775 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
386
+ 2022-04-09 03:20:19,785 INFO [decode.py:757] num_arcs after pruning: 6365
387
+ 2022-04-09 03:21:29,481 INFO [decode.py:736] Caught exception:
388
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.42 GiB already allocated; 1.12 GiB free; 29.27 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
389
+
390
+ 2022-04-09 03:21:29,481 INFO [decode.py:743] num_arcs before pruning: 872088
391
+ 2022-04-09 03:21:29,481 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
392
+ 2022-04-09 03:21:29,492 INFO [decode.py:757] num_arcs after pruning: 10043
393
+ 2022-04-09 03:22:01,760 INFO [decode_test.py:497] batch 4800/?, cuts processed until now is 14079
394
+ 2022-04-09 03:24:10,370 INFO [decode_test.py:497] batch 4900/?, cuts processed until now is 14298
395
+ 2022-04-09 03:26:10,811 INFO [decode_test.py:497] batch 5000/?, cuts processed until now is 14515
396
+ 2022-04-09 03:27:46,191 INFO [decode.py:736] Caught exception:
397
+
398
+ Some bad things happened. Please read the above error messages and stack
399
+ trace. If you are using Python, the following command may be helpful:
400
+
401
+ gdb --args python /path/to/your/code.py
402
+
403
+ (You can use `gdb` to debug the code. Please consider compiling
404
+ a debug version of k2.).
405
+
406
+ If you are unable to fix it, please open an issue at:
407
+
408
+ https://github.com/k2-fsa/k2/issues/new
409
+
410
+
411
+ 2022-04-09 03:27:46,192 INFO [decode.py:743] num_arcs before pruning: 246382
412
+ 2022-04-09 03:27:46,192 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
413
+ 2022-04-09 03:27:46,253 INFO [decode.py:757] num_arcs after pruning: 6775
414
+ 2022-04-09 03:28:15,199 INFO [decode_test.py:497] batch 5100/?, cuts processed until now is 14718
415
+ 2022-04-09 03:29:19,807 INFO [decode.py:736] Caught exception:
416
+ CUDA out of memory. Tried to allocate 6.15 GiB (GPU 0; 31.75 GiB total capacity; 26.67 GiB already allocated; 1.11 GiB free; 29.29 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
417
+
418
+ 2022-04-09 03:29:19,808 INFO [decode.py:743] num_arcs before pruning: 220820
419
+ 2022-04-09 03:29:19,808 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
420
+ 2022-04-09 03:29:19,815 INFO [decode.py:757] num_arcs after pruning: 13482
421
+ 2022-04-09 03:30:16,045 INFO [decode_test.py:497] batch 5200/?, cuts processed until now is 14930
422
+ 2022-04-09 03:32:12,235 INFO [decode_test.py:497] batch 5300/?, cuts processed until now is 15128
423
+ 2022-04-09 03:33:06,358 INFO [decode.py:736] Caught exception:
424
+
425
+ Some bad things happened. Please read the above error messages and stack
426
+ trace. If you are using Python, the following command may be helpful:
427
+
428
+ gdb --args python /path/to/your/code.py
429
+
430
+ (You can use `gdb` to debug the code. Please consider compiling
431
+ a debug version of k2.).
432
+
433
+ If you are unable to fix it, please open an issue at:
434
+
435
+ https://github.com/k2-fsa/k2/issues/new
436
+
437
+
438
+ 2022-04-09 03:33:06,359 INFO [decode.py:743] num_arcs before pruning: 190203
439
+ 2022-04-09 03:33:06,359 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
440
+ 2022-04-09 03:33:06,413 INFO [decode.py:757] num_arcs after pruning: 6202
441
+ 2022-04-09 03:34:14,862 INFO [decode_test.py:497] batch 5400/?, cuts processed until now is 15327
442
+ 2022-04-09 03:36:18,973 INFO [decode_test.py:497] batch 5500/?, cuts processed until now is 15531
443
+ 2022-04-09 03:38:18,633 INFO [decode_test.py:497] batch 5600/?, cuts processed until now is 15724
444
+ 2022-04-09 03:38:48,490 INFO [decode.py:736] Caught exception:
445
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.52 GiB already allocated; 3.07 GiB free; 27.32 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
446
+
447
+ 2022-04-09 03:38:48,491 INFO [decode.py:743] num_arcs before pruning: 554330
448
+ 2022-04-09 03:38:48,491 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
449
+ 2022-04-09 03:38:48,500 INFO [decode.py:757] num_arcs after pruning: 10730
450
+ 2022-04-09 03:39:51,281 INFO [decode.py:736] Caught exception:
451
+ CUDA out of memory. Tried to allocate 4.83 GiB (GPU 0; 31.75 GiB total capacity; 25.96 GiB already allocated; 1.31 GiB free; 29.08 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
452
+
453
+ 2022-04-09 03:39:51,281 INFO [decode.py:743] num_arcs before pruning: 160031
454
+ 2022-04-09 03:39:51,281 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
455
+ 2022-04-09 03:39:51,288 INFO [decode.py:757] num_arcs after pruning: 4270
456
+ 2022-04-09 03:40:28,016 INFO [decode_test.py:497] batch 5700/?, cuts processed until now is 15908
457
+ 2022-04-09 03:40:46,608 INFO [decode.py:736] Caught exception:
458
+ CUDA out of memory. Tried to allocate 2.58 GiB (GPU 0; 31.75 GiB total capacity; 27.28 GiB already allocated; 1.32 GiB free; 29.07 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
459
+
460
+ 2022-04-09 03:40:46,608 INFO [decode.py:743] num_arcs before pruning: 406026
461
+ 2022-04-09 03:40:46,608 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
462
+ 2022-04-09 03:40:46,616 INFO [decode.py:757] num_arcs after pruning: 11179
463
+ 2022-04-09 03:42:16,464 INFO [decode.py:736] Caught exception:
464
+ CUDA out of memory. Tried to allocate 2.29 GiB (GPU 0; 31.75 GiB total capacity; 26.71 GiB already allocated; 1.32 GiB free; 29.07 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
465
+
466
+ 2022-04-09 03:42:16,464 INFO [decode.py:743] num_arcs before pruning: 639824
467
+ 2022-04-09 03:42:16,464 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
468
+ 2022-04-09 03:42:16,476 INFO [decode.py:757] num_arcs after pruning: 5520
469
+ 2022-04-09 03:42:52,683 INFO [decode_test.py:497] batch 5800/?, cuts processed until now is 16094
470
+ 2022-04-09 03:44:51,754 INFO [decode_test.py:497] batch 5900/?, cuts processed until now is 16289
471
+ 2022-04-09 03:46:52,121 INFO [decode_test.py:497] batch 6000/?, cuts processed until now is 16488
472
+ 2022-04-09 03:48:54,739 INFO [decode_test.py:497] batch 6100/?, cuts processed until now is 16661
473
+ 2022-04-09 03:49:24,829 INFO [decode.py:736] Caught exception:
474
+ CUDA out of memory. Tried to allocate 1.84 GiB (GPU 0; 31.75 GiB total capacity; 28.87 GiB already allocated; 409.75 MiB free; 29.99 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
475
+
476
+ 2022-04-09 03:49:24,830 INFO [decode.py:743] num_arcs before pruning: 443401
477
+ 2022-04-09 03:49:24,830 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
478
+ 2022-04-09 03:49:24,837 INFO [decode.py:757] num_arcs after pruning: 5211
479
+ 2022-04-09 03:50:27,492 INFO [decode.py:736] Caught exception:
480
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.35 GiB already allocated; 2.15 GiB free; 28.24 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
481
+
482
+ 2022-04-09 03:50:27,493 INFO [decode.py:743] num_arcs before pruning: 361598
483
+ 2022-04-09 03:50:27,493 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
484
+ 2022-04-09 03:50:27,507 INFO [decode.py:757] num_arcs after pruning: 8660
485
+ 2022-04-09 03:51:02,856 INFO [decode_test.py:497] batch 6200/?, cuts processed until now is 16828
486
+ 2022-04-09 03:53:03,912 INFO [decode_test.py:497] batch 6300/?, cuts processed until now is 17002
487
+ 2022-04-09 03:55:04,964 INFO [decode_test.py:497] batch 6400/?, cuts processed until now is 17181
488
+ 2022-04-09 03:55:08,345 INFO [decode.py:736] Caught exception:
489
+ CUDA out of memory. Tried to allocate 4.89 GiB (GPU 0; 31.75 GiB total capacity; 26.28 GiB already allocated; 2.16 GiB free; 28.24 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
490
+
491
+ 2022-04-09 03:55:08,345 INFO [decode.py:743] num_arcs before pruning: 867262
492
+ 2022-04-09 03:55:08,345 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
493
+ 2022-04-09 03:55:08,356 INFO [decode.py:757] num_arcs after pruning: 6494
494
+ 2022-04-09 03:56:03,884 INFO [decode.py:736] Caught exception:
495
+ CUDA out of memory. Tried to allocate 1.90 GiB (GPU 0; 31.75 GiB total capacity; 28.97 GiB already allocated; 1.16 GiB free; 29.23 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
496
+
497
+ 2022-04-09 03:56:03,885 INFO [decode.py:743] num_arcs before pruning: 233755
498
+ 2022-04-09 03:56:03,885 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
499
+ 2022-04-09 03:56:03,910 INFO [decode.py:757] num_arcs after pruning: 5823
500
+ 2022-04-09 03:57:08,774 INFO [decode_test.py:497] batch 6500/?, cuts processed until now is 17347
501
+ 2022-04-09 03:59:01,245 INFO [decode_test.py:497] batch 6600/?, cuts processed until now is 17502
502
+ 2022-04-09 03:59:13,147 INFO [decode.py:736] Caught exception:
503
+ CUDA out of memory. Tried to allocate 5.80 GiB (GPU 0; 31.75 GiB total capacity; 26.73 GiB already allocated; 1.17 GiB free; 29.22 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
504
+
505
+ 2022-04-09 03:59:13,147 INFO [decode.py:743] num_arcs before pruning: 174004
506
+ 2022-04-09 03:59:13,147 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
507
+ 2022-04-09 03:59:13,155 INFO [decode.py:757] num_arcs after pruning: 6857
508
+ 2022-04-09 04:00:59,687 INFO [decode_test.py:497] batch 6700/?, cuts processed until now is 17661
509
+ 2022-04-09 04:03:01,660 INFO [decode_test.py:497] batch 6800/?, cuts processed until now is 17823
510
+ 2022-04-09 04:04:55,219 INFO [decode_test.py:497] batch 6900/?, cuts processed until now is 17997
511
+ 2022-04-09 04:07:05,841 INFO [decode_test.py:497] batch 7000/?, cuts processed until now is 18159
512
+ 2022-04-09 04:09:04,994 INFO [decode_test.py:497] batch 7100/?, cuts processed until now is 18299
513
+ 2022-04-09 04:11:07,439 INFO [decode_test.py:497] batch 7200/?, cuts processed until now is 18432
514
+ 2022-04-09 04:13:18,126 INFO [decode_test.py:497] batch 7300/?, cuts processed until now is 18552
515
+ 2022-04-09 04:15:23,102 INFO [decode_test.py:497] batch 7400/?, cuts processed until now is 18656
516
+ 2022-04-09 04:17:49,550 INFO [decode_test.py:497] batch 7500/?, cuts processed until now is 18798
517
+ 2022-04-09 04:19:16,128 INFO [decode.py:736] Caught exception:
518
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.34 GiB already allocated; 2.12 GiB free; 28.27 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
519
+
520
+ 2022-04-09 04:19:16,129 INFO [decode.py:743] num_arcs before pruning: 1155990
521
+ 2022-04-09 04:19:16,129 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
522
+ 2022-04-09 04:19:16,143 INFO [decode.py:757] num_arcs after pruning: 9141
523
+ 2022-04-09 04:20:19,961 INFO [decode_test.py:497] batch 7600/?, cuts processed until now is 18945
524
+ 2022-04-09 04:22:44,642 INFO [decode_test.py:497] batch 7700/?, cuts processed until now is 19084
525
+ 2022-04-09 04:23:18,184 INFO [decode.py:841] Caught exception:
526
+ CUDA out of memory. Tried to allocate 1.26 GiB (GPU 0; 31.75 GiB total capacity; 27.36 GiB already allocated; 881.75 MiB free; 29.53 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
527
+
528
+ 2022-04-09 04:23:18,184 INFO [decode.py:843] num_paths before decreasing: 1000
529
+ 2022-04-09 04:23:18,184 INFO [decode.py:852] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
530
+ 2022-04-09 04:23:18,184 INFO [decode.py:858] num_paths after decreasing: 500
531
+ 2022-04-09 04:24:52,959 INFO [decode.py:736] Caught exception:
532
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.53 GiB already allocated; 2.12 GiB free; 28.27 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
533
+
534
+ 2022-04-09 04:24:52,960 INFO [decode.py:743] num_arcs before pruning: 624026
535
+ 2022-04-09 04:24:52,960 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
536
+ 2022-04-09 04:24:52,972 INFO [decode.py:757] num_arcs after pruning: 10008
537
+ 2022-04-09 04:25:07,718 INFO [decode_test.py:497] batch 7800/?, cuts processed until now is 19232
538
+ 2022-04-09 04:25:31,876 INFO [decode.py:736] Caught exception:
539
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.51 GiB already allocated; 2.12 GiB free; 28.27 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
540
+
541
+ 2022-04-09 04:25:31,876 INFO [decode.py:743] num_arcs before pruning: 688909
542
+ 2022-04-09 04:25:31,877 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
543
+ 2022-04-09 04:25:31,887 INFO [decode.py:757] num_arcs after pruning: 8886
544
+ 2022-04-09 04:25:57,970 INFO [decode.py:736] Caught exception:
545
+ CUDA out of memory. Tried to allocate 5.04 GiB (GPU 0; 31.75 GiB total capacity; 25.95 GiB already allocated; 2.12 GiB free; 28.27 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
546
+
547
+ 2022-04-09 04:25:57,971 INFO [decode.py:743] num_arcs before pruning: 891176
548
+ 2022-04-09 04:25:57,971 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
549
+ 2022-04-09 04:25:57,982 INFO [decode.py:757] num_arcs after pruning: 10106
550
+ 2022-04-09 04:26:19,609 INFO [decode.py:736] Caught exception:
551
+ CUDA out of memory. Tried to allocate 2.63 GiB (GPU 0; 31.75 GiB total capacity; 27.60 GiB already allocated; 327.75 MiB free; 30.07 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
552
+
553
+ 2022-04-09 04:26:19,609 INFO [decode.py:743] num_arcs before pruning: 415376
554
+ 2022-04-09 04:26:19,609 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
555
+ 2022-04-09 04:26:19,620 INFO [decode.py:757] num_arcs after pruning: 7771
556
+ 2022-04-09 04:27:33,059 INFO [decode_test.py:497] batch 7900/?, cuts processed until now is 19375
557
+ 2022-04-09 04:29:43,649 INFO [decode_test.py:497] batch 8000/?, cuts processed until now is 19510
558
+ 2022-04-09 04:30:20,590 INFO [decode.py:736] Caught exception:
559
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.65 GiB already allocated; 2.12 GiB free; 28.27 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
560
+
561
+ 2022-04-09 04:30:20,591 INFO [decode.py:743] num_arcs before pruning: 330767
562
+ 2022-04-09 04:30:20,591 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
563
+ 2022-04-09 04:30:20,606 INFO [decode.py:757] num_arcs after pruning: 5820
564
+ 2022-04-09 04:31:55,818 INFO [decode_test.py:497] batch 8100/?, cuts processed until now is 19643
565
+ 2022-04-09 04:34:11,720 INFO [decode_test.py:497] batch 8200/?, cuts processed until now is 19776
566
+ 2022-04-09 04:35:04,147 INFO [decode.py:736] Caught exception:
567
+ CUDA out of memory. Tried to allocate 4.49 GiB (GPU 0; 31.75 GiB total capacity; 24.38 GiB already allocated; 2.12 GiB free; 28.27 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
568
+
569
+ 2022-04-09 04:35:04,147 INFO [decode.py:743] num_arcs before pruning: 533967
570
+ 2022-04-09 04:35:04,147 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
571
+ 2022-04-09 04:35:04,157 INFO [decode.py:757] num_arcs after pruning: 3449
572
+ 2022-04-09 04:36:15,595 INFO [decode.py:736] Caught exception:
573
+ CUDA out of memory. Tried to allocate 8.00 GiB (GPU 0; 31.75 GiB total capacity; 19.67 GiB already allocated; 2.12 GiB free; 28.27 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
574
+
575
+ 2022-04-09 04:36:15,595 INFO [decode.py:743] num_arcs before pruning: 397138
576
+ 2022-04-09 04:36:15,596 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
577
+ 2022-04-09 04:36:15,605 INFO [decode.py:757] num_arcs after pruning: 6775
578
+ 2022-04-09 04:36:31,844 INFO [decode_test.py:497] batch 8300/?, cuts processed until now is 19882
579
+ 2022-04-09 04:37:04,130 INFO [decode.py:736] Caught exception:
580
+
581
+ Some bad things happened. Please read the above error messages and stack
582
+ trace. If you are using Python, the following command may be helpful:
583
+
584
+ gdb --args python /path/to/your/code.py
585
+
586
+ (You can use `gdb` to debug the code. Please consider compiling
587
+ a debug version of k2.).
588
+
589
+ If you are unable to fix it, please open an issue at:
590
+
591
+ https://github.com/k2-fsa/k2/issues/new
592
+
593
+
594
+ 2022-04-09 04:37:04,130 INFO [decode.py:743] num_arcs before pruning: 456591
595
+ 2022-04-09 04:37:04,130 INFO [decode.py:746] This OOM is not an error. You can ignore it. If your model does not converge well, or --max-duration is too large, or the input sound file is difficult to decode, you will meet this exception.
596
+ 2022-04-09 04:37:04,180 INFO [decode.py:757] num_arcs after pruning: 5275
597
+ 2022-04-09 04:57:33,432 INFO [decode_test.py:567]
598
+ For test, WER of different settings are:
599
+ ngram_lm_scale_0.3_attention_scale_0.7 10.58 best for test
600
+ ngram_lm_scale_0.5_attention_scale_1.3 10.58
601
+ ngram_lm_scale_0.3_attention_scale_0.5 10.59
602
+ ngram_lm_scale_0.3_attention_scale_0.6 10.59
603
+ ngram_lm_scale_0.3_attention_scale_0.9 10.59
604
+ ngram_lm_scale_0.3_attention_scale_1.0 10.59
605
+ ngram_lm_scale_0.3_attention_scale_1.1 10.59
606
+ ngram_lm_scale_0.3_attention_scale_1.2 10.59
607
+ ngram_lm_scale_0.3_attention_scale_1.3 10.59
608
+ ngram_lm_scale_0.5_attention_scale_1.0 10.59
609
+ ngram_lm_scale_0.5_attention_scale_1.1 10.59
610
+ ngram_lm_scale_0.5_attention_scale_1.2 10.59
611
+ ngram_lm_scale_0.5_attention_scale_1.5 10.59
612
+ ngram_lm_scale_0.5_attention_scale_1.7 10.59
613
+ ngram_lm_scale_0.5_attention_scale_1.9 10.59
614
+ ngram_lm_scale_0.5_attention_scale_2.0 10.59
615
+ ngram_lm_scale_0.5_attention_scale_2.1 10.59
616
+ ngram_lm_scale_0.5_attention_scale_2.2 10.59
617
+ ngram_lm_scale_0.5_attention_scale_2.3 10.59
618
+ ngram_lm_scale_0.6_attention_scale_1.9 10.59
619
+ ngram_lm_scale_0.6_attention_scale_2.0 10.59
620
+ ngram_lm_scale_0.6_attention_scale_2.1 10.59
621
+ ngram_lm_scale_0.6_attention_scale_2.2 10.59
622
+ ngram_lm_scale_0.6_attention_scale_2.3 10.59
623
+ ngram_lm_scale_0.6_attention_scale_2.5 10.59
624
+ ngram_lm_scale_0.3_attention_scale_1.5 10.6
625
+ ngram_lm_scale_0.3_attention_scale_1.7 10.6
626
+ ngram_lm_scale_0.3_attention_scale_1.9 10.6
627
+ ngram_lm_scale_0.3_attention_scale_2.0 10.6
628
+ ngram_lm_scale_0.3_attention_scale_2.1 10.6
629
+ ngram_lm_scale_0.3_attention_scale_2.2 10.6
630
+ ngram_lm_scale_0.3_attention_scale_2.3 10.6
631
+ ngram_lm_scale_0.3_attention_scale_2.5 10.6
632
+ ngram_lm_scale_0.5_attention_scale_0.9 10.6
633
+ ngram_lm_scale_0.5_attention_scale_2.5 10.6
634
+ ngram_lm_scale_0.5_attention_scale_3.0 10.6
635
+ ngram_lm_scale_0.6_attention_scale_1.3 10.6
636
+ ngram_lm_scale_0.6_attention_scale_1.5 10.6
637
+ ngram_lm_scale_0.6_attention_scale_1.7 10.6
638
+ ngram_lm_scale_0.6_attention_scale_3.0 10.6
639
+ ngram_lm_scale_0.3_attention_scale_0.3 10.61
640
+ ngram_lm_scale_0.3_attention_scale_3.0 10.61
641
+ ngram_lm_scale_0.5_attention_scale_4.0 10.61
642
+ ngram_lm_scale_0.5_attention_scale_5.0 10.61
643
+ ngram_lm_scale_0.6_attention_scale_1.2 10.61
644
+ ngram_lm_scale_0.6_attention_scale_4.0 10.61
645
+ ngram_lm_scale_0.6_attention_scale_5.0 10.61
646
+ ngram_lm_scale_0.7_attention_scale_1.7 10.61
647
+ ngram_lm_scale_0.7_attention_scale_1.9 10.61
648
+ ngram_lm_scale_0.7_attention_scale_2.0 10.61
649
+ ngram_lm_scale_0.7_attention_scale_2.1 10.61
650
+ ngram_lm_scale_0.7_attention_scale_2.2 10.61
651
+ ngram_lm_scale_0.7_attention_scale_2.3 10.61
652
+ ngram_lm_scale_0.7_attention_scale_2.5 10.61
653
+ ngram_lm_scale_0.7_attention_scale_3.0 10.61
654
+ ngram_lm_scale_0.7_attention_scale_4.0 10.61
655
+ ngram_lm_scale_0.7_attention_scale_5.0 10.61
656
+ ngram_lm_scale_0.1_attention_scale_1.1 10.62
657
+ ngram_lm_scale_0.3_attention_scale_4.0 10.62
658
+ ngram_lm_scale_0.3_attention_scale_5.0 10.62
659
+ ngram_lm_scale_0.5_attention_scale_0.7 10.62
660
+ ngram_lm_scale_0.6_attention_scale_1.0 10.62
661
+ ngram_lm_scale_0.6_attention_scale_1.1 10.62
662
+ ngram_lm_scale_0.7_attention_scale_1.5 10.62
663
+ ngram_lm_scale_0.9_attention_scale_3.0 10.62
664
+ ngram_lm_scale_0.9_attention_scale_4.0 10.62
665
+ ngram_lm_scale_0.9_attention_scale_5.0 10.62
666
+ ngram_lm_scale_1.0_attention_scale_4.0 10.62
667
+ ngram_lm_scale_1.1_attention_scale_5.0 10.62
668
+ ngram_lm_scale_0.05_attention_scale_1.1 10.63
669
+ ngram_lm_scale_0.05_attention_scale_1.2 10.63
670
+ ngram_lm_scale_0.08_attention_scale_0.9 10.63
671
+ ngram_lm_scale_0.08_attention_scale_1.0 10.63
672
+ ngram_lm_scale_0.08_attention_scale_1.1 10.63
673
+ ngram_lm_scale_0.08_attention_scale_1.2 10.63
674
+ ngram_lm_scale_0.08_attention_scale_1.3 10.63
675
+ ngram_lm_scale_0.08_attention_scale_1.9 10.63
676
+ ngram_lm_scale_0.08_attention_scale_2.0 10.63
677
+ ngram_lm_scale_0.08_attention_scale_2.1 10.63
678
+ ngram_lm_scale_0.08_attention_scale_2.2 10.63
679
+ ngram_lm_scale_0.08_attention_scale_2.3 10.63
680
+ ngram_lm_scale_0.08_attention_scale_3.0 10.63
681
+ ngram_lm_scale_0.1_attention_scale_0.5 10.63
682
+ ngram_lm_scale_0.1_attention_scale_0.6 10.63
683
+ ngram_lm_scale_0.1_attention_scale_0.7 10.63
684
+ ngram_lm_scale_0.1_attention_scale_0.9 10.63
685
+ ngram_lm_scale_0.1_attention_scale_1.0 10.63
686
+ ngram_lm_scale_0.1_attention_scale_1.2 10.63
687
+ ngram_lm_scale_0.1_attention_scale_1.3 10.63
688
+ ngram_lm_scale_0.1_attention_scale_1.7 10.63
689
+ ngram_lm_scale_0.1_attention_scale_1.9 10.63
690
+ ngram_lm_scale_0.1_attention_scale_2.0 10.63
691
+ ngram_lm_scale_0.1_attention_scale_2.1 10.63
692
+ ngram_lm_scale_0.1_attention_scale_2.2 10.63
693
+ ngram_lm_scale_0.1_attention_scale_2.3 10.63
694
+ ngram_lm_scale_0.1_attention_scale_2.5 10.63
695
+ ngram_lm_scale_0.1_attention_scale_3.0 10.63
696
+ ngram_lm_scale_0.1_attention_scale_5.0 10.63
697
+ ngram_lm_scale_0.5_attention_scale_0.6 10.63
698
+ ngram_lm_scale_0.6_attention_scale_0.9 10.63
699
+ ngram_lm_scale_0.9_attention_scale_2.3 10.63
700
+ ngram_lm_scale_0.9_attention_scale_2.5 10.63
701
+ ngram_lm_scale_1.0_attention_scale_5.0 10.63
702
+ ngram_lm_scale_1.2_attention_scale_5.0 10.63
703
+ ngram_lm_scale_0.01_attention_scale_0.9 10.64
704
+ ngram_lm_scale_0.01_attention_scale_1.0 10.64
705
+ ngram_lm_scale_0.01_attention_scale_1.1 10.64
706
+ ngram_lm_scale_0.01_attention_scale_1.2 10.64
707
+ ngram_lm_scale_0.01_attention_scale_4.0 10.64
708
+ ngram_lm_scale_0.01_attention_scale_5.0 10.64
709
+ ngram_lm_scale_0.05_attention_scale_0.5 10.64
710
+ ngram_lm_scale_0.05_attention_scale_0.6 10.64
711
+ ngram_lm_scale_0.05_attention_scale_0.7 10.64
712
+ ngram_lm_scale_0.05_attention_scale_0.9 10.64
713
+ ngram_lm_scale_0.05_attention_scale_1.0 10.64
714
+ ngram_lm_scale_0.05_attention_scale_1.3 10.64
715
+ ngram_lm_scale_0.05_attention_scale_1.5 10.64
716
+ ngram_lm_scale_0.05_attention_scale_1.7 10.64
717
+ ngram_lm_scale_0.05_attention_scale_1.9 10.64
718
+ ngram_lm_scale_0.05_attention_scale_2.0 10.64
719
+ ngram_lm_scale_0.05_attention_scale_2.1 10.64
720
+ ngram_lm_scale_0.05_attention_scale_2.2 10.64
721
+ ngram_lm_scale_0.05_attention_scale_2.3 10.64
722
+ ngram_lm_scale_0.05_attention_scale_2.5 10.64
723
+ ngram_lm_scale_0.05_attention_scale_3.0 10.64
724
+ ngram_lm_scale_0.05_attention_scale_4.0 10.64
725
+ ngram_lm_scale_0.05_attention_scale_5.0 10.64
726
+ ngram_lm_scale_0.08_attention_scale_0.5 10.64
727
+ ngram_lm_scale_0.08_attention_scale_0.6 10.64
728
+ ngram_lm_scale_0.08_attention_scale_0.7 10.64
729
+ ngram_lm_scale_0.08_attention_scale_1.5 10.64
730
+ ngram_lm_scale_0.08_attention_scale_1.7 10.64
731
+ ngram_lm_scale_0.08_attention_scale_2.5 10.64
732
+ ngram_lm_scale_0.08_attention_scale_4.0 10.64
733
+ ngram_lm_scale_0.08_attention_scale_5.0 10.64
734
+ ngram_lm_scale_0.1_attention_scale_0.3 10.64
735
+ ngram_lm_scale_0.1_attention_scale_1.5 10.64
736
+ ngram_lm_scale_0.1_attention_scale_4.0 10.64
737
+ ngram_lm_scale_0.7_attention_scale_1.3 10.64
738
+ ngram_lm_scale_0.9_attention_scale_2.2 10.64
739
+ ngram_lm_scale_1.0_attention_scale_3.0 10.64
740
+ ngram_lm_scale_1.1_attention_scale_4.0 10.64
741
+ ngram_lm_scale_1.3_attention_scale_5.0 10.64
742
+ ngram_lm_scale_0.01_attention_scale_0.6 10.65
743
+ ngram_lm_scale_0.01_attention_scale_0.7 10.65
744
+ ngram_lm_scale_0.01_attention_scale_1.3 10.65
745
+ ngram_lm_scale_0.01_attention_scale_1.5 10.65
746
+ ngram_lm_scale_0.01_attention_scale_1.7 10.65
747
+ ngram_lm_scale_0.01_attention_scale_1.9 10.65
748
+ ngram_lm_scale_0.01_attention_scale_2.0 10.65
749
+ ngram_lm_scale_0.01_attention_scale_2.1 10.65
750
+ ngram_lm_scale_0.01_attention_scale_2.2 10.65
751
+ ngram_lm_scale_0.01_attention_scale_2.3 10.65
752
+ ngram_lm_scale_0.01_attention_scale_2.5 10.65
753
+ ngram_lm_scale_0.01_attention_scale_3.0 10.65
754
+ ngram_lm_scale_0.08_attention_scale_0.3 10.65
755
+ ngram_lm_scale_0.5_attention_scale_0.5 10.65
756
+ ngram_lm_scale_0.6_attention_scale_0.7 10.65
757
+ ngram_lm_scale_0.7_attention_scale_1.1 10.65
758
+ ngram_lm_scale_0.7_attention_scale_1.2 10.65
759
+ ngram_lm_scale_0.9_attention_scale_2.1 10.65
760
+ ngram_lm_scale_1.2_attention_scale_4.0 10.65
761
+ ngram_lm_scale_0.05_attention_scale_0.3 10.66
762
+ ngram_lm_scale_0.7_attention_scale_1.0 10.66
763
+ ngram_lm_scale_0.9_attention_scale_1.9 10.66
764
+ ngram_lm_scale_0.9_attention_scale_2.0 10.66
765
+ ngram_lm_scale_1.0_attention_scale_2.5 10.66
766
+ ngram_lm_scale_1.1_attention_scale_3.0 10.66
767
+ ngram_lm_scale_0.01_attention_scale_0.5 10.67
768
+ ngram_lm_scale_0.1_attention_scale_0.08 10.67
769
+ ngram_lm_scale_0.1_attention_scale_0.1 10.67
770
+ ngram_lm_scale_0.6_attention_scale_0.6 10.67
771
+ ngram_lm_scale_0.9_attention_scale_1.7 10.67
772
+ ngram_lm_scale_1.0_attention_scale_2.2 10.67
773
+ ngram_lm_scale_1.0_attention_scale_2.3 10.67
774
+ ngram_lm_scale_1.3_attention_scale_4.0 10.67
775
+ ngram_lm_scale_1.5_attention_scale_5.0 10.67
776
+ ngram_lm_scale_0.01_attention_scale_0.3 10.68
777
+ ngram_lm_scale_0.08_attention_scale_0.08 10.68
778
+ ngram_lm_scale_0.08_attention_scale_0.1 10.68
779
+ ngram_lm_scale_0.3_attention_scale_0.08 10.68
780
+ ngram_lm_scale_0.3_attention_scale_0.1 10.68
781
+ ngram_lm_scale_0.7_attention_scale_0.9 10.68
782
+ ngram_lm_scale_1.0_attention_scale_2.0 10.68
783
+ ngram_lm_scale_1.0_attention_scale_2.1 10.68
784
+ ngram_lm_scale_1.1_attention_scale_2.5 10.68
785
+ ngram_lm_scale_1.2_attention_scale_3.0 10.68
786
+ ngram_lm_scale_0.1_attention_scale_0.05 10.69
787
+ ngram_lm_scale_0.5_attention_scale_0.3 10.69
788
+ ngram_lm_scale_0.9_attention_scale_1.5 10.69
789
+ ngram_lm_scale_1.0_attention_scale_1.9 10.69
790
+ ngram_lm_scale_1.1_attention_scale_2.3 10.69
791
+ ngram_lm_scale_0.05_attention_scale_0.1 10.7
792
+ ngram_lm_scale_0.08_attention_scale_0.05 10.7
793
+ ngram_lm_scale_0.3_attention_scale_0.05 10.7
794
+ ngram_lm_scale_0.6_attention_scale_0.5 10.7
795
+ ngram_lm_scale_1.1_attention_scale_2.2 10.7
796
+ ngram_lm_scale_1.5_attention_scale_4.0 10.7
797
+ ngram_lm_scale_1.7_attention_scale_5.0 10.7
798
+ ngram_lm_scale_0.05_attention_scale_0.08 10.71
799
+ ngram_lm_scale_1.1_attention_scale_2.1 10.71
800
+ ngram_lm_scale_1.2_attention_scale_2.5 10.71
801
+ ngram_lm_scale_1.3_attention_scale_3.0 10.71
802
+ ngram_lm_scale_0.01_attention_scale_0.1 10.72
803
+ ngram_lm_scale_0.05_attention_scale_0.05 10.72
804
+ ngram_lm_scale_0.08_attention_scale_0.01 10.72
805
+ ngram_lm_scale_0.1_attention_scale_0.01 10.72
806
+ ngram_lm_scale_0.3_attention_scale_0.01 10.72
807
+ ngram_lm_scale_0.7_attention_scale_0.7 10.72
808
+ ngram_lm_scale_0.9_attention_scale_1.3 10.72
809
+ ngram_lm_scale_1.0_attention_scale_1.7 10.72
810
+ ngram_lm_scale_1.1_attention_scale_2.0 10.72
811
+ ngram_lm_scale_0.01_attention_scale_0.08 10.73
812
+ ngram_lm_scale_0.9_attention_scale_1.2 10.73
813
+ ngram_lm_scale_1.1_attention_scale_1.9 10.73
814
+ ngram_lm_scale_1.2_attention_scale_2.3 10.73
815
+ ngram_lm_scale_1.0_attention_scale_1.5 10.74
816
+ ngram_lm_scale_1.2_attention_scale_2.2 10.74
817
+ ngram_lm_scale_1.3_attention_scale_2.5 10.74
818
+ ngram_lm_scale_1.9_attention_scale_5.0 10.74
819
+ ngram_lm_scale_0.01_attention_scale_0.05 10.75
820
+ ngram_lm_scale_0.05_attention_scale_0.01 10.75
821
+ ngram_lm_scale_0.7_attention_scale_0.6 10.75
822
+ ngram_lm_scale_0.9_attention_scale_1.1 10.75
823
+ ngram_lm_scale_1.1_attention_scale_1.7 10.75
824
+ ngram_lm_scale_1.2_attention_scale_2.1 10.75
825
+ ngram_lm_scale_1.7_attention_scale_4.0 10.75
826
+ ngram_lm_scale_1.2_attention_scale_2.0 10.76
827
+ ngram_lm_scale_1.3_attention_scale_2.3 10.76
828
+ ngram_lm_scale_2.0_attention_scale_5.0 10.76
829
+ ngram_lm_scale_1.0_attention_scale_1.3 10.77
830
+ ngram_lm_scale_1.2_attention_scale_1.9 10.77
831
+ ngram_lm_scale_1.5_attention_scale_3.0 10.77
832
+ ngram_lm_scale_0.01_attention_scale_0.01 10.78
833
+ ngram_lm_scale_0.6_attention_scale_0.3 10.78
834
+ ngram_lm_scale_0.7_attention_scale_0.5 10.78
835
+ ngram_lm_scale_0.9_attention_scale_1.0 10.78
836
+ ngram_lm_scale_2.1_attention_scale_5.0 10.78
837
+ ngram_lm_scale_1.1_attention_scale_1.5 10.79
838
+ ngram_lm_scale_1.3_attention_scale_2.2 10.79
839
+ ngram_lm_scale_0.5_attention_scale_0.1 10.8
840
+ ngram_lm_scale_1.0_attention_scale_1.2 10.8
841
+ ngram_lm_scale_1.3_attention_scale_2.1 10.8
842
+ ngram_lm_scale_1.9_attention_scale_4.0 10.8
843
+ ngram_lm_scale_2.2_attention_scale_5.0 10.8
844
+ ngram_lm_scale_0.5_attention_scale_0.08 10.81
845
+ ngram_lm_scale_0.9_attention_scale_0.9 10.81
846
+ ngram_lm_scale_1.2_attention_scale_1.7 10.81
847
+ ngram_lm_scale_1.3_attention_scale_2.0 10.81
848
+ ngram_lm_scale_1.0_attention_scale_1.1 10.82
849
+ ngram_lm_scale_0.5_attention_scale_0.05 10.83
850
+ ngram_lm_scale_1.1_attention_scale_1.3 10.83
851
+ ngram_lm_scale_1.3_attention_scale_1.9 10.83
852
+ ngram_lm_scale_1.5_attention_scale_2.5 10.84
853
+ ngram_lm_scale_2.3_attention_scale_5.0 10.84
854
+ ngram_lm_scale_1.0_attention_scale_1.0 10.85
855
+ ngram_lm_scale_1.2_attention_scale_1.5 10.85
856
+ ngram_lm_scale_2.0_attention_scale_4.0 10.85
857
+ ngram_lm_scale_1.1_attention_scale_1.2 10.86
858
+ ngram_lm_scale_1.7_attention_scale_3.0 10.86
859
+ ngram_lm_scale_0.5_attention_scale_0.01 10.87
860
+ ngram_lm_scale_1.5_attention_scale_2.3 10.87
861
+ ngram_lm_scale_0.7_attention_scale_0.3 10.88
862
+ ngram_lm_scale_0.9_attention_scale_0.7 10.88
863
+ ngram_lm_scale_1.3_attention_scale_1.7 10.88
864
+ ngram_lm_scale_1.0_attention_scale_0.9 10.89
865
+ ngram_lm_scale_1.5_attention_scale_2.2 10.89
866
+ ngram_lm_scale_2.1_attention_scale_4.0 10.89
867
+ ngram_lm_scale_1.1_attention_scale_1.1 10.91
868
+ ngram_lm_scale_0.6_attention_scale_0.1 10.92
869
+ ngram_lm_scale_0.9_attention_scale_0.6 10.92
870
+ ngram_lm_scale_1.5_attention_scale_2.1 10.92
871
+ ngram_lm_scale_1.2_attention_scale_1.3 10.93
872
+ ngram_lm_scale_2.5_attention_scale_5.0 10.93
873
+ ngram_lm_scale_0.6_attention_scale_0.08 10.94
874
+ ngram_lm_scale_2.2_attention_scale_4.0 10.94
875
+ ngram_lm_scale_1.1_attention_scale_1.0 10.95
876
+ ngram_lm_scale_1.3_attention_scale_1.5 10.95
877
+ ngram_lm_scale_1.5_attention_scale_2.0 10.96
878
+ ngram_lm_scale_1.2_attention_scale_1.2 10.97
879
+ ngram_lm_scale_1.7_attention_scale_2.5 10.97
880
+ ngram_lm_scale_0.6_attention_scale_0.05 10.98
881
+ ngram_lm_scale_1.9_attention_scale_3.0 10.98
882
+ ngram_lm_scale_1.0_attention_scale_0.7 10.99
883
+ ngram_lm_scale_1.5_attention_scale_1.9 10.99
884
+ ngram_lm_scale_2.3_attention_scale_4.0 10.99
885
+ ngram_lm_scale_0.9_attention_scale_0.5 11.0
886
+ ngram_lm_scale_1.1_attention_scale_0.9 11.0
887
+ ngram_lm_scale_0.6_attention_scale_0.01 11.02
888
+ ngram_lm_scale_1.2_attention_scale_1.1 11.02
889
+ ngram_lm_scale_1.7_attention_scale_2.3 11.03
890
+ ngram_lm_scale_1.3_attention_scale_1.3 11.05
891
+ ngram_lm_scale_2.0_attention_scale_3.0 11.05
892
+ ngram_lm_scale_1.7_attention_scale_2.2 11.07
893
+ ngram_lm_scale_1.0_attention_scale_0.6 11.08
894
+ ngram_lm_scale_1.5_attention_scale_1.7 11.08
895
+ ngram_lm_scale_1.2_attention_scale_1.0 11.09
896
+ ngram_lm_scale_0.7_attention_scale_0.1 11.1
897
+ ngram_lm_scale_1.3_attention_scale_1.2 11.1
898
+ ngram_lm_scale_1.7_attention_scale_2.1 11.11
899
+ ngram_lm_scale_2.1_attention_scale_3.0 11.12
900
+ ngram_lm_scale_2.5_attention_scale_4.0 11.12
901
+ ngram_lm_scale_0.7_attention_scale_0.08 11.13
902
+ ngram_lm_scale_1.9_attention_scale_2.5 11.13
903
+ ngram_lm_scale_1.7_attention_scale_2.0 11.14
904
+ ngram_lm_scale_1.2_attention_scale_0.9 11.16
905
+ ngram_lm_scale_1.1_attention_scale_0.7 11.17
906
+ ngram_lm_scale_1.3_attention_scale_1.1 11.17
907
+ ngram_lm_scale_3.0_attention_scale_5.0 11.17
908
+ ngram_lm_scale_0.7_attention_scale_0.05 11.18
909
+ ngram_lm_scale_1.5_attention_scale_1.5 11.18
910
+ ngram_lm_scale_1.0_attention_scale_0.5 11.19
911
+ ngram_lm_scale_1.7_attention_scale_1.9 11.2
912
+ ngram_lm_scale_2.2_attention_scale_3.0 11.21
913
+ ngram_lm_scale_1.9_attention_scale_2.3 11.22
914
+ ngram_lm_scale_2.0_attention_scale_2.5 11.23
915
+ ngram_lm_scale_0.9_attention_scale_0.3 11.25
916
+ ngram_lm_scale_1.3_attention_scale_1.0 11.26
917
+ ngram_lm_scale_0.7_attention_scale_0.01 11.27
918
+ ngram_lm_scale_1.9_attention_scale_2.2 11.27
919
+ ngram_lm_scale_1.1_attention_scale_0.6 11.29
920
+ ngram_lm_scale_2.3_attention_scale_3.0 11.31
921
+ ngram_lm_scale_1.7_attention_scale_1.7 11.33
922
+ ngram_lm_scale_1.5_attention_scale_1.3 11.34
923
+ ngram_lm_scale_1.9_attention_scale_2.1 11.34
924
+ ngram_lm_scale_2.0_attention_scale_2.3 11.34
925
+ ngram_lm_scale_2.1_attention_scale_2.5 11.35
926
+ ngram_lm_scale_1.3_attention_scale_0.9 11.36
927
+ ngram_lm_scale_1.2_attention_scale_0.7 11.39
928
+ ngram_lm_scale_1.9_attention_scale_2.0 11.4
929
+ ngram_lm_scale_2.0_attention_scale_2.2 11.4
930
+ ngram_lm_scale_1.5_attention_scale_1.2 11.43
931
+ ngram_lm_scale_1.1_attention_scale_0.5 11.44
932
+ ngram_lm_scale_2.0_attention_scale_2.1 11.47
933
+ ngram_lm_scale_2.1_attention_scale_2.3 11.47
934
+ ngram_lm_scale_2.2_attention_scale_2.5 11.47
935
+ ngram_lm_scale_1.9_attention_scale_1.9 11.48
936
+ ngram_lm_scale_1.7_attention_scale_1.5 11.5
937
+ ngram_lm_scale_2.5_attention_scale_3.0 11.51
938
+ ngram_lm_scale_3.0_attention_scale_4.0 11.51
939
+ ngram_lm_scale_1.0_attention_scale_0.3 11.53
940
+ ngram_lm_scale_1.2_attention_scale_0.6 11.53
941
+ ngram_lm_scale_1.5_attention_scale_1.1 11.54
942
+ ngram_lm_scale_2.1_attention_scale_2.2 11.54
943
+ ngram_lm_scale_2.0_attention_scale_2.0 11.55
944
+ ngram_lm_scale_2.3_attention_scale_2.5 11.59
945
+ ngram_lm_scale_2.2_attention_scale_2.3 11.61
946
+ ngram_lm_scale_2.1_attention_scale_2.1 11.62
947
+ ngram_lm_scale_1.3_attention_scale_0.7 11.63
948
+ ngram_lm_scale_2.0_attention_scale_1.9 11.63
949
+ ngram_lm_scale_1.9_attention_scale_1.7 11.66
950
+ ngram_lm_scale_1.5_attention_scale_1.0 11.67
951
+ ngram_lm_scale_2.2_attention_scale_2.2 11.69
952
+ ngram_lm_scale_0.9_attention_scale_0.1 11.7
953
+ ngram_lm_scale_2.1_attention_scale_2.0 11.71
954
+ ngram_lm_scale_1.2_attention_scale_0.5 11.72
955
+ ngram_lm_scale_1.7_attention_scale_1.3 11.72
956
+ ngram_lm_scale_2.3_attention_scale_2.3 11.75
957
+ ngram_lm_scale_0.9_attention_scale_0.08 11.77
958
+ ngram_lm_scale_2.2_attention_scale_2.1 11.78
959
+ ngram_lm_scale_2.1_attention_scale_1.9 11.82
960
+ ngram_lm_scale_1.3_attention_scale_0.6 11.83
961
+ ngram_lm_scale_1.5_attention_scale_0.9 11.85
962
+ ngram_lm_scale_2.0_attention_scale_1.7 11.85
963
+ ngram_lm_scale_2.3_attention_scale_2.2 11.86
964
+ ngram_lm_scale_0.9_attention_scale_0.05 11.87
965
+ ngram_lm_scale_1.1_attention_scale_0.3 11.87
966
+ ngram_lm_scale_1.7_attention_scale_1.2 11.88
967
+ ngram_lm_scale_1.9_attention_scale_1.5 11.9
968
+ ngram_lm_scale_2.2_attention_scale_2.0 11.9
969
+ ngram_lm_scale_2.5_attention_scale_2.5 11.9
970
+ ngram_lm_scale_4.0_attention_scale_5.0 11.93
971
+ ngram_lm_scale_2.3_attention_scale_2.1 11.97
972
+ ngram_lm_scale_0.9_attention_scale_0.01 12.0
973
+ ngram_lm_scale_2.2_attention_scale_1.9 12.02
974
+ ngram_lm_scale_1.7_attention_scale_1.1 12.05
975
+ ngram_lm_scale_1.3_attention_scale_0.5 12.07
976
+ ngram_lm_scale_2.1_attention_scale_1.7 12.07
977
+ ngram_lm_scale_2.3_attention_scale_2.0 12.09
978
+ ngram_lm_scale_1.0_attention_scale_0.1 12.11
979
+ ngram_lm_scale_2.5_attention_scale_2.3 12.11
980
+ ngram_lm_scale_2.0_attention_scale_1.5 12.14
981
+ ngram_lm_scale_1.0_attention_scale_0.08 12.19
982
+ ngram_lm_scale_3.0_attention_scale_3.0 12.19
983
+ ngram_lm_scale_1.9_attention_scale_1.3 12.22
984
+ ngram_lm_scale_1.7_attention_scale_1.0 12.23
985
+ ngram_lm_scale_2.3_attention_scale_1.9 12.23
986
+ ngram_lm_scale_2.5_attention_scale_2.2 12.23
987
+ ngram_lm_scale_1.5_attention_scale_0.7 12.27
988
+ ngram_lm_scale_1.2_attention_scale_0.3 12.28
989
+ ngram_lm_scale_2.2_attention_scale_1.7 12.3
990
+ ngram_lm_scale_1.0_attention_scale_0.05 12.32
991
+ ngram_lm_scale_2.5_attention_scale_2.1 12.37
992
+ ngram_lm_scale_2.1_attention_scale_1.5 12.39
993
+ ngram_lm_scale_1.9_attention_scale_1.2 12.41
994
+ ngram_lm_scale_1.7_attention_scale_0.9 12.46
995
+ ngram_lm_scale_1.0_attention_scale_0.01 12.49
996
+ ngram_lm_scale_2.0_attention_scale_1.3 12.5
997
+ ngram_lm_scale_2.5_attention_scale_2.0 12.51
998
+ ngram_lm_scale_2.3_attention_scale_1.7 12.54
999
+ ngram_lm_scale_1.5_attention_scale_0.6 12.55
1000
+ ngram_lm_scale_1.1_attention_scale_0.1 12.58
1001
+ ngram_lm_scale_1.9_attention_scale_1.1 12.62
1002
+ ngram_lm_scale_2.2_attention_scale_1.5 12.64
1003
+ ngram_lm_scale_1.1_attention_scale_0.08 12.67
1004
+ ngram_lm_scale_2.5_attention_scale_1.9 12.67
1005
+ ngram_lm_scale_4.0_attention_scale_4.0 12.67
1006
+ ngram_lm_scale_1.3_attention_scale_0.3 12.71
1007
+ ngram_lm_scale_2.0_attention_scale_1.2 12.71
1008
+ ngram_lm_scale_2.1_attention_scale_1.3 12.78
1009
+ ngram_lm_scale_3.0_attention_scale_2.5 12.8
1010
+ ngram_lm_scale_1.1_attention_scale_0.05 12.81
1011
+ ngram_lm_scale_1.9_attention_scale_1.0 12.85
1012
+ ngram_lm_scale_1.5_attention_scale_0.5 12.86
1013
+ ngram_lm_scale_2.3_attention_scale_1.5 12.91
1014
+ ngram_lm_scale_2.0_attention_scale_1.1 12.92
1015
+ ngram_lm_scale_1.7_attention_scale_0.7 12.99
1016
+ ngram_lm_scale_2.1_attention_scale_1.2 12.99
1017
+ ngram_lm_scale_5.0_attention_scale_5.0 13.01
1018
+ ngram_lm_scale_1.1_attention_scale_0.01 13.02
1019
+ ngram_lm_scale_2.5_attention_scale_1.7 13.02
1020
+ ngram_lm_scale_2.2_attention_scale_1.3 13.05
1021
+ ngram_lm_scale_3.0_attention_scale_2.3 13.09
1022
+ ngram_lm_scale_1.2_attention_scale_0.1 13.1
1023
+ ngram_lm_scale_1.9_attention_scale_0.9 13.11
1024
+ ngram_lm_scale_2.0_attention_scale_1.0 13.17
1025
+ ngram_lm_scale_1.2_attention_scale_0.08 13.2
1026
+ ngram_lm_scale_2.1_attention_scale_1.1 13.22
1027
+ ngram_lm_scale_3.0_attention_scale_2.2 13.24
1028
+ ngram_lm_scale_2.2_attention_scale_1.2 13.28
1029
+ ngram_lm_scale_1.7_attention_scale_0.6 13.33
1030
+ ngram_lm_scale_2.3_attention_scale_1.3 13.34
1031
+ ngram_lm_scale_1.2_attention_scale_0.05 13.36
1032
+ ngram_lm_scale_3.0_attention_scale_2.1 13.42
1033
+ ngram_lm_scale_2.5_attention_scale_1.5 13.43
1034
+ ngram_lm_scale_2.0_attention_scale_0.9 13.48
1035
+ ngram_lm_scale_2.1_attention_scale_1.0 13.51
1036
+ ngram_lm_scale_2.2_attention_scale_1.1 13.56
1037
+ ngram_lm_scale_1.2_attention_scale_0.01 13.6
1038
+ ngram_lm_scale_2.3_attention_scale_1.2 13.6
1039
+ ngram_lm_scale_3.0_attention_scale_2.0 13.62
1040
+ ngram_lm_scale_1.3_attention_scale_0.1 13.65
1041
+ ngram_lm_scale_1.5_attention_scale_0.3 13.68
1042
+ ngram_lm_scale_1.7_attention_scale_0.5 13.72
1043
+ ngram_lm_scale_1.3_attention_scale_0.08 13.76
1044
+ ngram_lm_scale_1.9_attention_scale_0.7 13.78
1045
+ ngram_lm_scale_3.0_attention_scale_1.9 13.81
1046
+ ngram_lm_scale_2.1_attention_scale_0.9 13.82
1047
+ ngram_lm_scale_2.2_attention_scale_1.0 13.85
1048
+ ngram_lm_scale_4.0_attention_scale_3.0 13.85
1049
+ ngram_lm_scale_2.3_attention_scale_1.1 13.89
1050
+ ngram_lm_scale_1.3_attention_scale_0.05 13.94
1051
+ ngram_lm_scale_2.5_attention_scale_1.3 13.94
1052
+ ngram_lm_scale_5.0_attention_scale_4.0 13.97
1053
+ ngram_lm_scale_1.9_attention_scale_0.6 14.15
1054
+ ngram_lm_scale_2.0_attention_scale_0.7 14.16
1055
+ ngram_lm_scale_2.2_attention_scale_0.9 14.17
1056
+ ngram_lm_scale_2.3_attention_scale_1.0 14.19
1057
+ ngram_lm_scale_1.3_attention_scale_0.01 14.2
1058
+ ngram_lm_scale_2.5_attention_scale_1.2 14.2
1059
+ ngram_lm_scale_3.0_attention_scale_1.7 14.26
1060
+ ngram_lm_scale_2.5_attention_scale_1.1 14.48
1061
+ ngram_lm_scale_2.3_attention_scale_0.9 14.5
1062
+ ngram_lm_scale_2.1_attention_scale_0.7 14.53
1063
+ ngram_lm_scale_2.0_attention_scale_0.6 14.54
1064
+ ngram_lm_scale_1.9_attention_scale_0.5 14.57
1065
+ ngram_lm_scale_4.0_attention_scale_2.5 14.63
1066
+ ngram_lm_scale_1.7_attention_scale_0.3 14.64
1067
+ ngram_lm_scale_3.0_attention_scale_1.5 14.71
1068
+ ngram_lm_scale_1.5_attention_scale_0.1 14.75
1069
+ ngram_lm_scale_2.5_attention_scale_1.0 14.79
1070
+ ngram_lm_scale_2.2_attention_scale_0.7 14.86
1071
+ ngram_lm_scale_1.5_attention_scale_0.08 14.87
1072
+ ngram_lm_scale_2.1_attention_scale_0.6 14.91
1073
+ ngram_lm_scale_2.0_attention_scale_0.5 14.95
1074
+ ngram_lm_scale_4.0_attention_scale_2.3 14.98
1075
+ ngram_lm_scale_1.5_attention_scale_0.05 15.05
1076
+ ngram_lm_scale_2.5_attention_scale_0.9 15.12
1077
+ ngram_lm_scale_4.0_attention_scale_2.2 15.17
1078
+ ngram_lm_scale_2.3_attention_scale_0.7 15.21
1079
+ ngram_lm_scale_3.0_attention_scale_1.3 15.22
1080
+ ngram_lm_scale_2.2_attention_scale_0.6 15.27
1081
+ ngram_lm_scale_1.5_attention_scale_0.01 15.3
1082
+ ngram_lm_scale_5.0_attention_scale_3.0 15.32
1083
+ ngram_lm_scale_2.1_attention_scale_0.5 15.33
1084
+ ngram_lm_scale_4.0_attention_scale_2.1 15.37
1085
+ ngram_lm_scale_1.9_attention_scale_0.3 15.5
1086
+ ngram_lm_scale_3.0_attention_scale_1.2 15.51
1087
+ ngram_lm_scale_4.0_attention_scale_2.0 15.57
1088
+ ngram_lm_scale_2.3_attention_scale_0.6 15.61
1089
+ ngram_lm_scale_2.2_attention_scale_0.5 15.68
1090
+ ngram_lm_scale_1.7_attention_scale_0.1 15.72
1091
+ ngram_lm_scale_4.0_attention_scale_1.9 15.79
1092
+ ngram_lm_scale_3.0_attention_scale_1.1 15.82
1093
+ ngram_lm_scale_1.7_attention_scale_0.08 15.83
1094
+ ngram_lm_scale_2.5_attention_scale_0.7 15.85
1095
+ ngram_lm_scale_2.0_attention_scale_0.3 15.87
1096
+ ngram_lm_scale_2.3_attention_scale_0.5 16.0
1097
+ ngram_lm_scale_1.7_attention_scale_0.05 16.01
1098
+ ngram_lm_scale_3.0_attention_scale_1.0 16.11
1099
+ ngram_lm_scale_5.0_attention_scale_2.5 16.12
1100
+ ngram_lm_scale_2.5_attention_scale_0.6 16.19
1101
+ ngram_lm_scale_2.1_attention_scale_0.3 16.2
1102
+ ngram_lm_scale_4.0_attention_scale_1.7 16.22
1103
+ ngram_lm_scale_1.7_attention_scale_0.01 16.23
1104
+ ngram_lm_scale_3.0_attention_scale_0.9 16.4
1105
+ ngram_lm_scale_5.0_attention_scale_2.3 16.44
1106
+ ngram_lm_scale_1.9_attention_scale_0.1 16.5
1107
+ ngram_lm_scale_2.2_attention_scale_0.3 16.53
1108
+ ngram_lm_scale_2.5_attention_scale_0.5 16.54
1109
+ ngram_lm_scale_1.9_attention_scale_0.08 16.6
1110
+ ngram_lm_scale_5.0_attention_scale_2.2 16.6
1111
+ ngram_lm_scale_4.0_attention_scale_1.5 16.63
1112
+ ngram_lm_scale_1.9_attention_scale_0.05 16.74
1113
+ ngram_lm_scale_5.0_attention_scale_2.1 16.77
1114
+ ngram_lm_scale_2.3_attention_scale_0.3 16.81
1115
+ ngram_lm_scale_2.0_attention_scale_0.1 16.83
1116
+ ngram_lm_scale_2.0_attention_scale_0.08 16.92
1117
+ ngram_lm_scale_5.0_attention_scale_2.0 16.94
1118
+ ngram_lm_scale_1.9_attention_scale_0.01 16.95
1119
+ ngram_lm_scale_3.0_attention_scale_0.7 16.96
1120
+ ngram_lm_scale_2.0_attention_scale_0.05 17.05
1121
+ ngram_lm_scale_4.0_attention_scale_1.3 17.05
1122
+ ngram_lm_scale_2.1_attention_scale_0.1 17.11
1123
+ ngram_lm_scale_5.0_attention_scale_1.9 17.11
1124
+ ngram_lm_scale_2.1_attention_scale_0.08 17.21
1125
+ ngram_lm_scale_2.0_attention_scale_0.01 17.24
1126
+ ngram_lm_scale_3.0_attention_scale_0.6 17.26
1127
+ ngram_lm_scale_4.0_attention_scale_1.2 17.27
1128
+ ngram_lm_scale_2.5_attention_scale_0.3 17.28
1129
+ ngram_lm_scale_2.1_attention_scale_0.05 17.34
1130
+ ngram_lm_scale_2.2_attention_scale_0.1 17.38
1131
+ ngram_lm_scale_5.0_attention_scale_1.7 17.44
1132
+ ngram_lm_scale_2.2_attention_scale_0.08 17.46
1133
+ ngram_lm_scale_4.0_attention_scale_1.1 17.5
1134
+ ngram_lm_scale_2.1_attention_scale_0.01 17.52
1135
+ ngram_lm_scale_3.0_attention_scale_0.5 17.57
1136
+ ngram_lm_scale_2.2_attention_scale_0.05 17.59
1137
+ ngram_lm_scale_2.3_attention_scale_0.1 17.62
1138
+ ngram_lm_scale_2.3_attention_scale_0.08 17.7
1139
+ ngram_lm_scale_4.0_attention_scale_1.0 17.72
1140
+ ngram_lm_scale_2.2_attention_scale_0.01 17.76
1141
+ ngram_lm_scale_5.0_attention_scale_1.5 17.8
1142
+ ngram_lm_scale_2.3_attention_scale_0.05 17.82
1143
+ ngram_lm_scale_4.0_attention_scale_0.9 17.94
1144
+ ngram_lm_scale_2.3_attention_scale_0.01 17.98
1145
+ ngram_lm_scale_2.5_attention_scale_0.1 18.03
1146
+ ngram_lm_scale_2.5_attention_scale_0.08 18.1
1147
+ ngram_lm_scale_5.0_attention_scale_1.3 18.12
1148
+ ngram_lm_scale_3.0_attention_scale_0.3 18.17
1149
+ ngram_lm_scale_2.5_attention_scale_0.05 18.2
1150
+ ngram_lm_scale_5.0_attention_scale_1.2 18.29
1151
+ ngram_lm_scale_2.5_attention_scale_0.01 18.33
1152
+ ngram_lm_scale_4.0_attention_scale_0.7 18.36
1153
+ ngram_lm_scale_5.0_attention_scale_1.1 18.48
1154
+ ngram_lm_scale_4.0_attention_scale_0.6 18.58
1155
+ ngram_lm_scale_5.0_attention_scale_1.0 18.65
1156
+ ngram_lm_scale_3.0_attention_scale_0.1 18.75
1157
+ ngram_lm_scale_4.0_attention_scale_0.5 18.79
1158
+ ngram_lm_scale_3.0_attention_scale_0.08 18.81
1159
+ ngram_lm_scale_5.0_attention_scale_0.9 18.81
1160
+ ngram_lm_scale_3.0_attention_scale_0.05 18.89
1161
+ ngram_lm_scale_3.0_attention_scale_0.01 18.99
1162
+ ngram_lm_scale_5.0_attention_scale_0.7 19.11
1163
+ ngram_lm_scale_4.0_attention_scale_0.3 19.18
1164
+ ngram_lm_scale_5.0_attention_scale_0.6 19.25
1165
+ ngram_lm_scale_5.0_attention_scale_0.5 19.41
1166
+ ngram_lm_scale_4.0_attention_scale_0.1 19.57
1167
+ ngram_lm_scale_4.0_attention_scale_0.08 19.61
1168
+ ngram_lm_scale_4.0_attention_scale_0.05 19.67
1169
+ ngram_lm_scale_5.0_attention_scale_0.3 19.71
1170
+ ngram_lm_scale_4.0_attention_scale_0.01 19.73
1171
+ ngram_lm_scale_5.0_attention_scale_0.1 19.99
1172
+ ngram_lm_scale_5.0_attention_scale_0.08 20.01
1173
+ ngram_lm_scale_5.0_attention_scale_0.05 20.05
1174
+ ngram_lm_scale_5.0_attention_scale_0.01 20.11
1175
+
1176
+ 2022-04-09 04:57:33,455 INFO [decode_test.py:730] Done!