-
Hi,
Latest version of keras has no support for theano utils, is there any alternative for this:
Traceback (most recent call last):
File "cnn_blstm.py", line 28, in
from birnn import BiDirectio…
-
Hi
I employed the blstm-crf model for Chinese named entity recognition,it worked but results was no better than the one without crf, and I got real_path_score larger than all_path score,is that poss…
-
使用 Tensorflow 2.x 作为 keras 后端，报如下错误
```python
Traceback (most recent call last):
File ".\vuldeepecker.py", line 100, in
main()
File ".\vuldeepecker.py", line 96, in main
blstm.tra…
-
In the original LAS paper, we can read :
> In our model, we stack 3 pBLSTMs on top of the bottom BLSTM layer to reduce the time resolution
> 2
> 3 = 8 times. This allows the attention model (see …
-
I found that in the [origin CRNN implementation](https://github.com/bgshih/crnn/blob/master/model/crnn_demo/config.lua), the author used 2 layers of BLSTM, but i only found 2 layers of LSTM in your pr…
-
BLSTM可以自动标注，但我看你测试数据集和训练数据集都是标注好的，最后用一段话来测试。不应该是用测试集来测试吗？我有点没有看懂
-
Hi,
I was going through your repository. I could not find results of LSTM and BLSTM on the 2 speaker .wav (audio files) generated by you. Can you please add them?
Also, have you tried this algorit…
-
-
Hi @hirofumi0810 ,
I'm trying to replicate the results of the LAS model that you shared in this [table](https://github.com/hirofumi0810/neural_sp#csj-wer).
I'm using the default [run.sh](https:…
-
On some internal data, and maybe also Switchboard and Librispeech.
Using `nn.Conformer`. Making it somewhat more standard if possible, and then deviate from it when it makes sense.
Also compare …