Skip to content

fix dtype#672

Merged
Xreki merged 3 commits intoPaddlePaddle:developfrom
Honglei-Qiu:qhl_dtype
Mar 12, 2026
Merged

fix dtype#672
Xreki merged 3 commits intoPaddlePaddle:developfrom
Honglei-Qiu:qhl_dtype

Conversation

@Honglei-Qiu
Copy link
Contributor

PR Category

other

Description

@paddle-bot
Copy link

paddle-bot bot commented Mar 10, 2026

Thanks for your contribution!

rel_model_path: str,
traced_model: fx.GraphModule,
pass_name: str,
inputs: list = None,
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

这个inputs就不用再传进来了吧?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

已修改

# Load inputs from the updated meta files (dtype matches meta exactly),
# run ShapeProp to get real runtime dtypes, then prune redundant .to() nodes.
try:
_, meta_inputs = get_torch_module_and_inputs(str(output_dir))
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

在这之前调用一下empty_cache吧,清除一下前面执行申请的存储空间。

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

已修改

Copy link
Collaborator

@Xreki Xreki left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@Xreki Xreki closed this Mar 12, 2026
@Xreki Xreki reopened this Mar 12, 2026
@Xreki Xreki merged commit ee0eb99 into PaddlePaddle:develop Mar 12, 2026
5 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants