未验证 提交 e7efea71 编写于 作者: S SunAhong1993 提交者: GitHub

Merge pull request #553 from SunAhong1993/real_dev

for new version
__version__ = "1.0.1"
__version__ = "1.0.2"
from .core.program import PaddleGraph
......
......@@ -281,7 +281,7 @@ class PyTorchOpMapper(OpMapper):
scope_name=scope_name,
data=output_name)
if self.input_examples is not None:
input_np = self.input_examples[input_ct].detach().numpy()
input_np = self.input_examples[input_ct].cpu().detach().numpy()
self.inputs_info[
output_name] = [list(input_np.shape), str(input_np.dtype)]
return [], [output_name]
......
Markdown is supported
0% .
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