Skip to content

[BugFix] Fix DS/Kimi crash on non-first PP ranks when resolving input length #31752

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
ch-wan merged 2 commits into from
Jul 24, 2026
+8 −4
Merged

[BugFix] Fix DS/Kimi crash on non-first PP ranks when resolving input length #31752

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
dc56379
[BugFix] fix none-type error with pp_size > 1 for Ds/Kimi
ZhengWG Jul 20, 2026
1d38842
Merge branch 'main' into py/fix-dsv2-pp-git
ZhengWG Jul 20, 2026
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
12 changes: 8 additions & 4 deletions python/sglang/srt/models/deepseek_v2.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2842,10 +2842,14 @@ def forward(
input_embeds: torch.Tensor = None,
pp_proxy_tensors: Optional[PPProxyTensors] = None,
) -> torch.Tensor:
# Minor fix for multi-modal model: input_ids is None
len_input_ids = (
input_ids.shape[0] if input_ids is not None else input_embeds.shape[0]
)
# Multi-modal: input_ids may be None (use input_embeds).
# Non-first PP ranks: both are None (activations via pp_proxy_tensors).
if input_ids is not None:
len_input_ids = input_ids.shape[0]
elif input_embeds is not None:
len_input_ids = input_embeds.shape[0]
else:
len_input_ids = pp_proxy_tensors["hidden_states"].shape[0]
if self.dsa_enable_prefill_cp:
if can_dsa_cp_split(
len_input_ids, self.cp_size, self.use_dsa, forward_batch
Expand Down
Loading