From 70645e48072bf193fbf069949f1d2b10fddac8a3 Mon Sep 17 00:00:00 2001 From: pointerhacker <145901472+pointerhacker@users.noreply.github.com> Date: Tue, 15 Oct 2024 17:50:51 +0800 Subject: [PATCH] =?UTF-8?q?=E6=95=B0=E6=8D=AE=E5=B9=B6=E8=A1=8C=E5=8F=AF?= =?UTF-8?q?=E8=83=BD=E5=AF=BC=E8=87=B4=E7=9A=84=E6=A8=A1=E5=9E=8B=E8=AE=AD?= =?UTF-8?q?=E7=BB=83=E6=8A=A5=E9=94=99=20(#2139)?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit * fix: 修复数据并行训练中ä¼可能会出现的错误 * fix: 修复数据并行训练中ä¼可能会出现的错误 * fix: 修复数据并行ènot need tensor --------- Co-authored-by: zhaochaojin --- funasr/models/sense_voice/model.py | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/funasr/models/sense_voice/model.py b/funasr/models/sense_voice/model.py index 25e9fafd3..ba82091e9 100644 --- a/funasr/models/sense_voice/model.py +++ b/funasr/models/sense_voice/model.py @@ -555,7 +555,8 @@ class SenseVoiceEncoderSmall(nn.Module): ilens: torch.Tensor, ): """Embed positions in tensor.""" - masks = sequence_mask(ilens, device=ilens.device)[:, None, :] + maxlen = xs_pad.shape[1] + masks = sequence_mask(ilens, maxlen = maxlen, device=ilens.device)[:, None, :] xs_pad *= self.output_size() ** 0.5