Daniel Povey
6252282fd0
Add deriv-balancing code
2022-03-04 20:19:11 +08:00
Daniel Povey
eb3ed54202
Reduce scale from 50 to 20
2022-03-04 15:56:45 +08:00
Daniel Povey
9cc5999829
Fix duplicate Swish; replace norm+swish with swish+exp-scale in convolution module
2022-03-04 15:50:51 +08:00
Daniel Povey
7e88999641
Increase scale from 20 to 50.
2022-03-04 14:31:29 +08:00
Daniel Povey
3207bd98a9
Increase scale on Scale from 4 to 20
2022-03-04 13:16:40 +08:00
Daniel Povey
503f8d521c
Fix bug in diagnostics
2022-03-04 13:08:56 +08:00
Daniel Povey
3d9ddc2016
Fix backprop bug
2022-03-04 12:29:44 +08:00
Daniel Povey
cd216f50b6
Add import
2022-03-04 11:03:01 +08:00
Daniel Povey
bc6c720e25
Combine ExpScale and swish for memory reduction
2022-03-04 10:52:05 +08:00
Daniel Povey
23b3aa233c
Double learning rate of exp-scale units
2022-03-04 00:42:37 +08:00
Daniel Povey
5c177fc52b
pelu_base->expscale, add 2xExpScale in subsampling, and in feedforward units.
2022-03-03 23:52:03 +08:00
Daniel Povey
3fb559d2f0
Add baseline for the PeLU expt, keeping only the small normalization-related changes.
2022-03-02 18:27:08 +08:00
Daniel Povey
9ed7d55a84
Small bug fixes/imports
2022-03-02 16:34:55 +08:00
Daniel Povey
9d1b4ae046
Add pelu to this good-performing setup..
2022-03-02 16:33:27 +08:00
Daniel Povey
c1063def95
First version of rand-combine iterated-training-like idea.
2022-02-27 17:34:58 +08:00
Daniel Povey
63d8d935d4
Refactor/simplify ConformerEncoder
2022-02-27 13:56:15 +08:00
Daniel Povey
581786a6d3
Adding diagnostics code...
2022-02-27 13:44:43 +08:00
Daniel Povey
2af1b3af98
Remove ReLU in attention
2022-02-14 19:39:19 +08:00
Daniel Povey
d187ad8b73
Change max_frames from 0.2 to 0.15
2022-02-11 16:24:17 +08:00
Daniel Povey
4cd2c02fff
Fix num_time_masks code; revert 0.8 to 0.9
2022-02-10 15:53:11 +08:00
Daniel Povey
c170c53006
Change p=0.9 to p=0.8 in SpecAug
2022-02-10 14:59:14 +08:00
Daniel Povey
8aa50df4f0
Change p=0.5->0.9, mask_fraction 0.3->0.2
2022-02-09 22:52:53 +08:00
Daniel Povey
dd19a6a2b1
Fix to num_feature_masks bug I introduced; reduce max_frames_mask_fraction 0.4->0.3
2022-02-09 12:02:19 +08:00
Daniel Povey
bd36216e8c
Use much more aggressive SpecAug setup
2022-02-08 21:55:20 +08:00
Daniel Povey
beaf5bfbab
Merge specaug change from Mingshuang.
2022-02-08 19:42:23 +08:00
Daniel Povey
395065eb11
Merge branch 'spec-augment-change' of https://github.com/luomingshuang/icefall into attention_relu_specaug
2022-02-08 19:40:33 +08:00
Mingshuang Luo
3323cabf46
Experiments based on SpecAugment change
2022-02-08 14:25:31 +08:00
Daniel Povey
a859dcb205
Remove learnable offset, use relu instead.
2022-02-07 12:14:48 +08:00
Wei Kang
35ecd7e562
Fix torch.nn.Embedding error for torch below 1.8.0 ( #198 )
2022-02-06 21:59:54 +08:00
Daniel Povey
48a764eccf
Add min in q,k,v of attention
2022-02-06 21:19:37 +08:00
Daniel Povey
8f8ec223a7
Changes to fbank computation, use lilcom chunky writer
2022-02-06 21:18:40 +08:00
pkufool
fcd25bdfff
Fix torch.nn.Embedding error for torch below 1.8.0
2022-02-06 18:22:56 +08:00
Wei Kang
5ae80dfca7
Minor fixes ( #193 )
2022-01-27 18:01:17 +08:00
Piotr Żelasko
1731cc37bb
Black
2022-01-24 10:20:22 -05:00
Piotr Żelasko
f92c24a73a
Merge branch 'master' into feature/libri-conformer-phone-ctc
2022-01-24 10:18:56 -05:00
Piotr Żelasko
565c1d8413
Address code review
2022-01-24 10:17:47 -05:00
Piotr Żelasko
1d5fe8afa4
flake8
2022-01-21 17:27:02 -05:00
Piotr Żelasko
f0f35e6671
black
2022-01-21 17:22:41 -05:00
Piotr Żelasko
f28951f2b6
Add an assertion
2022-01-21 17:16:49 -05:00
Piotr Żelasko
3d109b121d
Remove train_phones.py and modify train.py instead
2022-01-21 17:08:53 -05:00
Fangjun Kuang
d6050eb02e
Fix calling optimized_transducer after new release. ( #182 )
2022-01-21 08:18:50 +08:00
Fangjun Kuang
f94ff19bfe
Refactor beam search and update results. ( #177 )
2022-01-18 16:40:19 +08:00
Fangjun Kuang
273e5fb2f3
Update git SHA1 for transducer_stateless model. ( #174 )
2022-01-10 11:58:17 +08:00
Fangjun Kuang
4c1b3665ee
Use optimized_transducer to compute transducer loss. ( #162 )
...
* WIP: Use optimized_transducer to compute transducer loss.
* Minor fixes.
* Fix decoding.
* Fix decoding.
* Add RESULTS.
* Update RESULTS.
* Update CI.
* Fix sampling rate for yesno recipe.
2022-01-10 11:54:58 +08:00
Fangjun Kuang
413b2e8569
Add git sha1 to RESULTS.md for conformer encoder + stateless decoder. ( #160 )
2021-12-28 12:04:01 +08:00
Fangjun Kuang
14c93add50
Remove batchnorm, weight decay, and SOS from transducer conformer encoder ( #155 )
...
* Remove batchnorm, weight decay, and SOS.
* Make --context-size configurable.
* Update results.
2021-12-27 16:01:10 +08:00
Fangjun Kuang
8187d6236c
Minor fix to maximum number of symbols per frame for RNN-T decoding. ( #157 )
...
* Minor fix to maximum number of symbols per frame RNN-T decoding.
* Minor fixes.
2021-12-24 21:48:40 +08:00
Fangjun Kuang
5b6699a835
Minor fixes to the RNN-T Conformer model ( #152 )
...
* Disable weight decay.
* Remove input feature batchnorm..
* Replace BatchNorm in the Conformer model with LayerNorm.
* Use tanh in the joint network.
* Remove sos ID.
* Reduce the number of decoder layers from 4 to 2.
* Minor fixes.
* Fix typos.
2021-12-23 13:54:25 +08:00
Fangjun Kuang
fb6a57e9e0
Increase the size of the context in the RNN-T decoder. ( #153 )
2021-12-23 07:55:02 +08:00
Fangjun Kuang
cb04c8a750
Limit the number of symbols per frame in RNN-T decoding. ( #151 )
2021-12-18 11:00:42 +08:00
