FUNASR训练
parent: b7b4a83c | 补丁 | 提交 | show whitespace
维石
2024-07-25 530125dca86b6cecb853ce08f2dfaab461fa762a 
bug fix
1个文件已修改
4 ■■■■■ 已修改文件
runtime/python/libtorch/funasr_torch/sensevoice_bin.py 4 ●●●●● 补丁 | 查看 | 原始文档 | blame | 历史 
 runtime/python/libtorch/funasr_torch/sensevoice_bin.py


@@ -167,8 +167,8 @@


            ctc_logits, encoder_out_lens = self.ort_infer(


                torch.Tensor(feats).to(self.device),


                torch.Tensor(feats_len).to(self.device),


                torch.tensor([_language_list]).to(self.device),


                torch.tensor([_textnorm_list]).to(self.device),


                torch.tensor(_language_list).to(self.device),


                torch.tensor(_textnorm_list).to(self.device),


            )


            # support batch_size=1 only currently


            x = ctc_logits[0, : encoder_out_lens[0].item(), :]