Skip to content

Make the default value of fsdp compatible with transformers less than 4.57 #7235

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
hjh0119 merged 22 commits into from
Dec 29, 2025
Merged

Make the default value of fsdp compatible with transformers less than 4.57 #7235

Changes from all commits
Commits
Show all changes
22 commits
Select commit Hold shift + click to select a range
923c87c
support hunyuanOCR
slin000111 Dec 15, 2025
277191e
Merge branch 'support_hunyuanvl' into main
slin000111 Dec 15, 2025
aa7a9e5
fix bug
slin000111 Dec 15, 2025
87b0da1
Merge branch 'support_hunyuanvl' into main
slin000111 Dec 15, 2025
2842e3d
fix typos
slin000111 Dec 15, 2025
9d35311
Merge branch 'support_hunyuanvl' into main
slin000111 Dec 15, 2025
48fb0a9
fix position_ids for multi images
slin000111 Dec 15, 2025
48d5fe5
Merge branch 'support_hunyuanvl' into main
slin000111 Dec 15, 2025
d78f66b
Merge branch 'modelscope:main' into main
slin000111 Dec 16, 2025
9a88ef8
Merge branch 'modelscope:main' into main
slin000111 Dec 17, 2025
78b1970
update faq
slin000111 Dec 22, 2025
9629222
Merge branch 'modelscope:main' into main
slin000111 Dec 22, 2025
7f8da0a
Merge branch 'update_faq' into main
slin000111 Dec 22, 2025
021418b
fix typos
slin000111 Dec 22, 2025
e095597
Merge branch 'update_faq' into main
slin000111 Dec 22, 2025
6e35ee3
Merge branch 'modelscope:main' into main
slin000111 Dec 24, 2025
195e160
update faq
slin000111 Dec 24, 2025
1a0c058
Merge branch 'modelscope:main' into main
slin000111 Dec 24, 2025
bcd1efb
Merge branch 'update_faq' into main
slin000111 Dec 24, 2025
3bb863e
fix link
slin000111 Dec 24, 2025
e3b0e55
Merge branch 'modelscope:main' into main
slin000111 Dec 29, 2025
00eaa14
Make the default value of fsdp compatible with previous versions of t…
slin000111 Dec 29, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions swift/llm/argument/train_args.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -290,6 +290,7 @@ def _init_deepspeed(self):

def _init_fsdp(self):
if not self.fsdp:
self.fsdp = []
Copy link
Copy Markdown
Contributor

@gemini-code-assist gemini-code-assist bot Dec 29, 2025

Choose a reason for hiding this comment

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

medium

While setting self.fsdp = [] fixes the compatibility issue, it introduces a type inconsistency. The fsdp attribute is type-hinted as Optional[str] on line 204, but you are assigning a list to it. To maintain type consistency, it would be better to set it to an empty string "". This is also a safe "disabled" value across different transformers versions and respects the established type hint.

Suggested change
self.fsdp = []
self.fsdp = ""

return

if is_mp() and not self.use_ray:
Expand Down