mirrors/icefall
1
0
Fork 0
mirror of https://github.com/k2-fsa/icefall.git synced 2025-08-14 12:32:20 +00:00
Code Issues Packages Projects Releases Wiki Activity
77 Commits 5 Branches 3 Tags
Commit Graph

77 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Daniel Povey
6636c05f12 Some configuration changes, trying to tune it so ctc_loss does not degrade from epoch 1.. 2021-09-23 19:38:57 +08:00
Daniel Povey
6fa0f16e0c Remove reconstruction loss, have randomly averaged CTC loss 2021-09-23 17:31:29 +08:00
Daniel Povey
3415dab779 Small code beautification 2021-09-23 11:39:25 +08:00
Daniel Povey
2213457bd3 Initially working version with delay_loss... 2021-09-23 11:25:42 +08:00
Daniel Povey
65b737576e train2.py not working due to issues in distributed training, hard to fix 2021-09-22 12:20:17 +08:00
Daniel Povey
6f8b7b9c3b First version that seems to be converging OK... 2021-09-21 21:52:17 +08:00
Daniel Povey
c4cc952265 Some configuration changes, change how prob_boost works 2021-09-21 12:06:41 +08:00
Daniel Povey
656de090bd Add some more debug stuff: seems like things move around too fast for negative branch to track.. 2021-09-20 16:11:30 +08:00
Daniel Povey
ed84795b47 Config changes, bug fix 2021-09-20 13:39:25 +08:00
Daniel Povey
2bad68a8ed Trying to figure out why it's not converging.. 2021-09-20 13:18:46 +08:00
Daniel Povey
39b6879d72 Version that is running... 2021-09-19 22:12:17 +08:00
Daniel Povey
3bad661f6f train.py draft.. 2021-09-19 21:18:12 +08:00
Daniel Povey
ef69661549 Change some defaults.. 2021-09-19 21:18:00 +08:00
Daniel Povey
b0dd4215fe Refactor so there is no bottleneck, only prediction 2021-09-19 15:38:34 +08:00
Daniel Povey
0f29f35a42 Changes to test, RE shifting.. 2021-09-18 23:04:50 +08:00
Daniel Povey
da3c9c7594 Some updates to tests, still figuring out issues.. 2021-09-18 21:47:31 +08:00
Daniel Povey
461cb7da6d Version that is successfully optimizing... 2021-09-18 16:40:55 +08:00
Daniel Povey
38081bc3e3 Some progress in testing.. 2021-09-18 15:00:27 +08:00
Daniel Povey
a20d490332 Get backward working 2021-09-18 12:36:50 +08:00
Daniel Povey
058fff0365 Get bidirectional conformer to run 2021-09-18 12:32:39 +08:00
Daniel Povey
a75f75bbad Fix bugs 2021-09-18 11:34:35 +08:00
Daniel Povey
c6c3750cab Testing configuration for conformer_ctc_bn 2021-09-17 18:55:34 +08:00
Daniel Povey
cfdfcf657d Initial drafts/work on bidirectional conformer 2021-09-17 13:47:54 +08:00
Daniel Povey
2b0370eb18 Copy conformer_ctc_bn scripts, no changes yet. 2021-09-15 11:42:59 +08:00
Daniel Povey
1d5e509261 Fix to madam.py, RE optimizer state 2021-09-14 13:13:48 +08:00
Daniel Povey
dfe773aa78 First version of conformer with discrete bottleneck 2021-09-10 18:51:16 +08:00
Daniel Povey
44b33b7f05 Init conformer_ctc_bn with copy of conformer_ctc files. 2021-09-10 16:13:24 +08:00
Daniel Povey
1078e4878c Add 1/sqrt(t) factor to gloam 2021-09-09 14:19:01 +08:00
Daniel Povey
c810e67342 Add some debugging code to train.py: 2021-09-09 14:03:04 +08:00
Fangjun Kuang
abadc71415
Use new APIs with k2.RaggedTensor (#38) 
* Use new APIs with k2.RaggedTensor

* Fix style issues.

* Update the installation doc, saying it requires at least k2 v1.7

* Use k2 v1.7
 2021-09-08 14:55:30 +08:00
Fangjun Kuang
331e5eb7ab
[doc] Fix typos. (#31) 2021-09-02 07:12:37 +08:00
Mingshuang Luo
5baa6a9f1c
fix a spelling mistake (tourch->touch) (#29) v1.0 2021-08-25 21:41:46 +08:00
Mingshuang Luo
eed3fc5610
Correct some spelling mistakes (#28) 
* Update index.rst (AS->ASR)

* Update conformer_ctc.rst (pretraind->pretrained)
 2021-08-25 17:48:34 +08:00
Fangjun Kuang
184dbb3ea5
Add documentation about code style and creating new recipes. (#27) 2021-08-25 14:48:41 +08:00
Fangjun Kuang
96e7f5c7ea
Release v0.1 (#26) v0.1 2021-08-24 21:30:30 +08:00
pkufool
f4223ee110
Add TDNN-LSTM-CTC Results (#25) 
* Add tdnn-lstm pretrained model and results

* Add docs for TDNN-LSTM-CTC

* Minor fix

* Fix typo

* Fix style checking
 2021-08-24 21:09:27 +08:00
Fangjun Kuang
1bd5dcc8ac
WIP: Add doc for the LibriSpeech recipe. (#24) 
* WIP: Add doc for the LibriSpeech recipe.

* Add more doc for LibriSpeech recipe.

* Add more doc for the LibriSpeech recipe.

* More doc.
 2021-08-24 20:28:32 +08:00
Fangjun Kuang
01da00dca0
WIP: Add documentation. (#22) 
* Begin to add documentation.

* WIP: Add documentation.

* Fix a typo.

* Add more doc for the recipe yesno.

* Add more doc for the yesno recipe.
 2021-08-24 14:28:08 +08:00
Fangjun Kuang
57cb611665
[yesno] Remove padding in TDNN (#21) 
* Disable SpecAug for yesno.

Also replace Adam with SGD.

* Remove padding in the model to make the results reproducible.
 2021-08-23 15:59:36 +08:00
Fangjun Kuang
6c2c9b9d74
Add recipe for the yes_no dataset. (#16) 
* Add recipe for the yes_no dataset.

* Refactoring: Remove unused code.

* Add Colab notebook for the yesno dataset.

* Add GitHub actions to run yesno.

* Fix a typo.

* Minor fixes.

* Train more epochs for GitHub actions.

* Minor fixes.

* Minor fixes.

* Fix style issues.
 2021-08-23 11:36:29 +08:00
pkufool
19c4214958
Fix code style and add copyright. (#18) 
* Fix style and add copyright

* Minor fix

* Remove duplicate lines

* Reformat conformer.py by black

* Reformat code style with black.

* Fix github workflows

* Fix lhotse installation

* Install icefall requirements

* Update k2 version, remove lhotse from test workflow
 2021-08-23 10:43:59 +08:00
Fangjun Kuang
8469f9ae0a
Refactor asr_datamodule. (#15) 
* WIP: Refactor asr_datamodule.

* Fixes after review.

* Minor fixes.
 2021-08-21 09:53:46 +08:00
Fangjun Kuang
0b656e4e1c
Add a link to Colab. (#14) 
It demonstrates the usages of pre-trained models.
 2021-08-20 15:43:25 +08:00
Fangjun Kuang
9d0cc9d829
Support computing nbest oracle WER. (#10) 
* Support computing nbest oracle WER.

* Add scale to all nbest based decoding/rescoring methods.

* Add script to run pretrained models.

* Use torchaudio to extract features.

* Support decoding multiple files at the same time.

Also, use kaldifeat for feature extraction.

* Support decoding with LM rescoring and attention-decoder rescoring.

* Minor fixes.

* Replace scale with lattice-score-scale.

* Add usage example with a provided pretrained model.
 2021-08-20 11:53:37 +08:00
pkufool
ef233486ae
The training script produce WER of 2.57% on librispeech test-clean (#13) 
* Add grad_clip and weight-decay, small fix of dataloader and masking

* Add RESULTS.md
 2021-08-20 10:08:08 +08:00
Fangjun Kuang
caa0b9e942
Fix an error in displaying decoding process. (#12) 2021-08-19 14:54:01 +08:00
Fangjun Kuang
1c3b13c7eb
Minor fixes. (#9) 2021-08-16 19:01:25 +08:00
Fangjun Kuang
12a2fd023e
Add doc about installation and usage (#7) 
* Add readme.

* Add TOC.

* fix typos

* Minor fixes after review.
 2021-08-12 12:44:04 +08:00
Fangjun Kuang
5a0b9bcb23
Refactoring (#4) 
* Fix an error in TDNN-LSTM training.

* WIP: Refactoring

* Refactor transformer.py

* Remove unused code.

* Minor fixes.
 2021-08-04 14:53:02 +08:00
Daniel Povey
cf8d76293d
Merge pull request #3 from csukuangfj/style-check 
Add CTC training
 2021-07-31 15:36:00 +08:00