Skip to content

fix: draft model revision misuse model revision #11893

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
zhyncs merged 3 commits into from
Nov 25, 2025
Merged

fix: draft model revision misuse model revision #11893

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
30daaf7
fix: draft model revision misuse model revision
gongwei-130 Oct 21, 2025
4bb6cb2
change speculative_draft_model_revision default to main
gongwei-130 Oct 21, 2025
e131fde
Merge branch 'main' into wei_fix_draft_model_revision
Qiaolin-Yu Nov 24, 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 python/sglang/srt/model_executor/model_runner.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -538,6 +538,7 @@ def initialize(self, min_per_gpu_memory: float):
draft_model_config = ModelConfig.from_server_args(
server_args,
model_path=(server_args.speculative_draft_model_path),
model_revision=server_args.speculative_draft_model_revision,
is_draft_model=True,
)

Expand Down
7 changes: 7 additions & 0 deletions python/sglang/srt/server_args.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1610,6 +1610,12 @@ def _handle_hicache(self):
)

def _handle_speculative_decoding(self):
if (
self.speculative_draft_model_path is not None
and self.speculative_draft_model_revision is None
):
self.speculative_draft_model_revision = "main"
Copy link
Copy Markdown
Collaborator

@Qiaolin-Yu Qiaolin-Yu Oct 22, 2025 •
edited
Loading

Choose a reason for hiding this comment

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

Why we need to set this by default?

Copy link
Copy Markdown
Collaborator Author

@gongwei-130 gongwei-130 Nov 19, 2025 •
edited
Loading

Choose a reason for hiding this comment

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

@Qiaolin-Yu sorry, lost track of this.

If we don't set a default value for it, it would default to None and pass in here, then from_server_args would use server_args.revision as draft model revision, which introduce problem.


if self.speculative_algorithm == "NEXTN":
self.speculative_algorithm = "EAGLE"

Expand Down Expand Up @@ -1658,6 +1664,7 @@ def _handle_speculative_decoding(self):
]:
if self.speculative_draft_model_path is None:
self.speculative_draft_model_path = self.model_path
self.speculative_draft_model_revision = self.revision
else:
logger.warning(
"DeepSeek MTP does not require setting speculative_draft_model_path."
Expand Down
Loading