mirrors/DeepSpeech
2
0
Fork 0
mirror of https://github.com/mozilla/DeepSpeech.git synced 2025-10-26 11:19:39 +00:00
Code Issues Packages Projects Releases Wiki Activity
1,504 Commits 63 Branches 109 Tags 17 GiB
augment
Commit Graph

147 Commits

Author SHA1 Message Date
nicolaspanel
7a10ed6971 allow custom export_batch_size (see #1920) 2019-03-01 15:10:05 +01:00
Reuben Morais
718ef951ad Update .compute script for new cluster setup 2019-02-19 13:26:51 -03:00
Alexandre Lissy
e6cca187d6 Enable TFLite post-training quantization 
Fixes #1850
 2019-02-14 16:41:09 +01:00
Reuben Morais
dd6938f7af Fix handling of Alphabet around evaluate.py 2018-12-15 09:59:49 -02:00
Reuben Morais
8ebfe80dd4
Set graph's random seed when training 2018-12-11 21:49:19 -02:00
Reuben Morais
c33b0d29d0
Merge pull request #1732 from mozilla/issue1505 
Disable StepCounterHook to avoid useless warning during validation (Fixes #1505)
 2018-11-17 16:07:34 -02:00
Vamsi Krishna Surapureddi
fcee7afe57 Fix #1731 is_chief reference from Config 2018-11-17 22:43:56 +05:30
Reuben Morais
af0620040a Disable StepCounterHook to avoid useless warning during validation 2018-11-17 14:03:30 -02:00
Reuben Morais
5cb1aff531 Rename config singleton from C to Config 2018-11-11 18:03:52 -02:00
Reuben Morais
a5bcecbe40 Address more review comments 2018-11-11 15:24:31 -02:00
Reuben Morais
c65c22fe31 Move globals handling out of util/coordinator.py 2018-11-09 21:08:03 -02:00
Reuben Morais
56dc024d29 Centralize WER report code into evaluate.py, call it from DeepSpeech.py 2018-11-09 00:26:50 -02:00
Reuben Morais
eee92c232d
Merge pull request #1693 from mozilla/ctcdecode-python 
Expose ctcdecode to Python and use it in evaluate.py
 2018-11-08 10:28:54 -02:00
Alexandre Lissy
d6642da05b Fix typo exporting the wrong tensor for TFLite model 2018-11-07 16:52:57 +01:00
Reuben Morais
a9e1ffc0fc Use ctcdecode bindings in evaluate.py 2018-11-03 12:07:54 -03:00
Reuben Morais
bb4551caa9 Extend Python Alphabet with config file and decode method 2018-11-02 14:00:11 -03:00
Alexandre Lissy
92ded598fb Call toco during export 2018-10-31 22:38:52 +01:00
Alexandre Lissy
5d30afdbad Training and directly exporting as TF Lite 2018-10-30 19:17:25 +01:00
Reuben Morais
3cc9b3711d Use ctcdecode in native client 2018-10-25 17:01:08 -03:00
Reuben Morais
46d1cece4f Remove initialize_from_frozen_model flag and support code 2018-10-23 15:08:38 -03:00
lissyx
55fb2305e2
Merge pull request #1611 from daanzu/pr-docs 
Fix docstrings that were being missed by Sphinx
 2018-10-15 11:35:40 +02:00
kdavis-mozilla
c38dc099d3 Fixed #1638 (Update Hyperparameters for 0.3.0) 2018-10-12 17:26:32 +02:00
daanzu
4dd5396571 Fix errors with uninitialized update_progressbar() field 
There are multiple paths that could lead to trying to read
update_progressbar.pbar without ever calling
update_progressbar(), but they already check for None.
 2018-10-04 08:03:38 -04:00
Alexandre Lissy
d1d559355a Remove WarpCTC 2018-10-03 10:19:25 +02:00
Alexandre Lissy
1b8b46ec95 Fix one shot inference 
Fixes #1612
 2018-10-02 14:20:45 +02:00
daanzu
06d4379994 Fix docstrings that were being missed by Sphinx 2018-10-01 10:06:39 -04:00
Reuben Morais
009f9b6c19 Retrigger v0.2.0 2018-09-18 13:40:18 -03:00
Reuben Morais
3663bca7e0 Update hyperparams from trained model 2018-09-17 23:55:40 -03:00
Reuben Morais
778f5deb9d Fix export graph 2018-09-14 11:42:42 -03:00
Reuben Morais
44e502e236 Feature caching 2018-09-14 11:14:36 -03:00
Reuben Morais
5ff3b31bbd Finish progressbar after final job is complete 2018-09-11 19:45:12 -03:00
Reuben Morais
25de412c9d Pass reuse flag to LSTMBlockFusedCell 2018-08-15 15:59:41 -03:00
Reuben Morais
38609d58d8 Wrap stdout in training progress bars 2018-08-15 09:48:25 -03:00
Bhavani Subramanian
f7fef57e64 Add support for specifying inter/intra-op parallelism threads. 2018-08-11 23:05:49 -03:00
Reuben Morais
295fa7128b Fix training progress bar on repeated epochs of the same set 2018-08-10 23:05:14 -03:00
George Fedoseev
d9997104b1 added progressbar 2018-08-10 12:14:34 +03:00
Reuben Morais
51cac1031b Address review comments 2018-08-03 14:46:05 -03:00
Reuben Morais
8fecc33333 Architecture changes from training 2018-08-02 13:22:24 -03:00
Reuben Morais
385bc7a65f Remove word count bonus in favor of length normalization 2018-08-02 13:22:24 -03:00
Reuben Morais
cbb33c8563 Finalize training graph before starting training 2018-08-02 13:22:24 -03:00
Reuben Morais
48778fcd0e Don't handle dropout rates as Python values 2018-08-02 13:22:24 -03:00
Reuben Morais
0b95ed6981 Switch to LSTMBlockFusedCell 2018-08-02 13:22:24 -03:00
Reuben Morais
831ebf4081 Add evaluate.py script to create a WER report from a checkpoint using the inference graph 2018-08-02 13:22:24 -03:00
Reuben Morais
400c9c0e5d Streaming inference 2018-08-02 13:22:24 -03:00
Alexandre Lissy
7a613c5392 Fix training from frozen graph 
Fixes #1472
 2018-07-24 13:32:25 +02:00
Reuben Morais
6ef186cfb0
Merge pull request #1391 from mozilla/log_step_loss 
Log step avg loss unconditionally
 2018-07-17 13:47:45 +00:00
Reuben Morais
da97317d5b Log step avg loss unconditionally 2018-06-15 10:49:14 -07:00
Reuben Morais
f25c89982e Don't apply dropout in evaluation steps 2018-06-15 10:36:40 -07:00
Reuben Morais
73ef114417 Disable checkpoint saving in MonitoredTrainingSession 2018-03-26 19:14:14 -03:00
Reuben Morais
983e13f218 Remove TF 1.0 compatibility hack 2018-02-14 15:27:39 +01:00