Skip to content

D
DeepSpeech

PaddlePaddle / DeepSpeech
大约 2 年 前同步成功

通知 210
Star 8425
Fork 1598
D
DeepSpeech
提交 c4a5ae38 编写于 作者: H Hui Zhang
浏览文件
操作

eliminate mul

上级 b7388ce2
隐藏空白更改
内联 并排
Showing with 2 addition and 2 deletion
paddlespeech/audio/utils/tensor_utils.py
浏览文件 @ c4a5ae38
...@@ -152,8 +152,8 @@ def add_sos_eos(ys_pad: paddle.Tensor, sos: int, eos: int, ...@@ -152,8 +152,8 @@ def add_sos_eos(ys_pad: paddle.Tensor, sos: int, eos: int,
# return pad_sequence(ys_in, padding_value=eos).transpose([1,0]), pad_sequence(ys_out, padding_value=ignore_id).transpose([1,0]) # return pad_sequence(ys_in, padding_value=eos).transpose([1,0]), pad_sequence(ys_out, padding_value=ignore_id).transpose([1,0])
B = ys_pad.shape[0] B = ys_pad.shape[0]
_sos = paddle.ones([B, 1], dtype=ys_pad.dtype) * sos _sos = paddle.full([B, 1], sos, dtype=ys_pad.dtype)
_eos = paddle.ones([B, 1], dtype=ys_pad.dtype) * eos _eos = paddle.full([B, 1], eos, dtype=ys_pad.dtype)
ys_in = paddle.cat([_sos, ys_pad], dim=1) ys_in = paddle.cat([_sos, ys_pad], dim=1)
mask_pad = (ys_in == ignore_id) mask_pad = (ys_in == ignore_id)
ys_in = ys_in.masked_fill(mask_pad, eos) ys_in = ys_in.masked_fill(mask_pad, eos)
......
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或