Skip to content

[CPU] fix no attribute 'can_fuse_mlp_allreduce' error #8010

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 1 commit into from
Jul 14, 2025
Merged

[CPU] fix no attribute 'can_fuse_mlp_allreduce' error #8010

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
8 changes: 5 additions & 3 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 @@ -462,7 +462,7 @@ def forward_normal(
if hasattr(self, "shared_experts") and use_intel_amx_backend(
self.shared_experts.gate_up_proj
):
return self.forward_cpu(hidden_states)
return self.forward_cpu(hidden_states, can_fuse_mlp_allreduce)

shared_output = self._forward_shared_experts(hidden_states)
# router_logits: (num_tokens, n_experts)
Expand All @@ -479,7 +479,9 @@ def forward_normal(
final_hidden_states = tensor_model_parallel_all_reduce(final_hidden_states)
return final_hidden_states

def forward_cpu(self, hidden_states: torch.Tensor) -> torch.Tensor:
def forward_cpu(
self, hidden_states: torch.Tensor, can_fuse_mlp_allreduce: bool = False
) -> torch.Tensor:
# router_logits: (num_tokens, n_experts)
router_logits = self.gate(hidden_states)
fused_experts_out = self.experts(
Expand Down Expand Up @@ -528,7 +530,7 @@ def forward_cpu(self, hidden_states: torch.Tensor) -> torch.Tensor:
None, # a2_scale
True, # is_vnni
)
if self.tp_size > 1 and not self.can_fuse_mlp_allreduce:
if self.tp_size > 1 and not can_fuse_mlp_allreduce:
final_hidden_states = tensor_model_parallel_all_reduce(final_hidden_states)
return final_hidden_states

Expand Down
Loading