speech_recognition/FunASR
1
0
Fork 0
mirror of https://github.com/modelscope/FunASR synced 2025-09-15 14:48:36 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

fix bug for contextual train

Browse Source
This commit is contained in:
shixian.shi 2023-11-24 09:46:51 +08:00
parent 269554388c
commit 34b2682fba
1 changed files with 1 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
funasr/models/e2e_asr_contextual_paraformer.py
View File

@ -207,7 +207,7 @@ class NeatContextualParaformer(Paraformer):
# 2b. Attention decoder branch # 2b. Attention decoder branch
if self.ctc_weight != 1.0: if self.ctc_weight != 1.0:
loss_att, acc_att, cer_att, wer_att, loss_pre, loss_ideal = self._calc_att_clas_loss( loss_att, acc_att, cer_att, wer_att, loss_pre, loss_ideal = self._calc_att_clas_loss(
encoder_out, encoder_out_lens, text, text_lengths, hotword_pad, hotword_lengths, ideal_attn encoder_out, encoder_out_lens, text, text_lengths, hotword_pad, hotword_lengths
) )
# 3. CTC-Att loss definition # 3. CTC-Att loss definition
@ -242,7 +242,6 @@ class NeatContextualParaformer(Paraformer):
ys_pad_lens: torch.Tensor, ys_pad_lens: torch.Tensor,
hotword_pad: torch.Tensor, hotword_pad: torch.Tensor,
hotword_lengths: torch.Tensor, hotword_lengths: torch.Tensor,
ideal_attn: torch.Tensor,
): ):
encoder_out_mask = (~make_pad_mask(encoder_out_lens, maxlen=encoder_out.size(1))[:, None, :]).to( encoder_out_mask = (~make_pad_mask(encoder_out_lens, maxlen=encoder_out.size(1))[:, None, :]).to(
encoder_out.device) encoder_out.device)