Thanks to visit codestin.com
Credit goes to github.com

Skip to content

[MoE] Fix misuse of num_experts as expert parallel group size (ep_size) #7551

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
tohtana merged 1 commit into from
Sep 10, 2025
Merged

[MoE] Fix misuse of num_experts as expert parallel group size (ep_size) #7551

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
4 changes: 3 additions & 1 deletion deepspeed/inference/engine.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -125,7 +125,7 @@ def __init__(self, model, config):
moe = False

if moe and dist.get_world_size() > 1:
self._create_ep_parallel_group(config.moe.moe_experts)
self._create_ep_parallel_group(config.moe.ep_size)

# We only support three modes: 1) user specified policy for tensor-parallelism, 2) kernel injection (replace_with_kernel_inject), and 3) automatic tensor parallelism if tp_size > 1.
if self.injection_dict:
Expand Down Expand Up @@ -267,6 +267,8 @@ def _create_ep_parallel_group(self, moe_experts):
self.expert_mp_group.update({e: None})
for moe_ep_size in self.ep_group.keys():
num_ep_groups = dist.get_world_size() // moe_ep_size
if num_ep_groups == 0:
raise ValueError(f"Invalid ep_size={moe_ep_size} for world_size={dist.get_world_size()}")
for i in range(num_ep_groups):
ep_cnt = i * moe_ep_size
size = dist.get_world_size() if moe_ep_size > dist.get_world_size() else moe_ep_size
Expand Down
Loading