Skip to content

Fix - moe lora support #1

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

Open
yushengsu-thu wants to merge 2 commits into
base: add-moe-lora-support
Choose a base branch
from
Open

Fix - moe lora support #1

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
a8512d5
fix problem in triton op backend
yushengsu-thu Jan 28, 2026
f88666d
update
yushengsu-thu Jan 28, 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
20 changes: 14 additions & 6 deletions python/sglang/srt/lora/mem_pool.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -169,15 +169,23 @@ def get_lora_A_shape(
"num_local_experts",
getattr(self.base_hf_config, "num_experts", 0),
)
# Allocate all MoE buffers with the same maximum rank dimension
# to ensure consistent kernel compilation. The maximum stacking factor is 2.
max_rank_dim = (
max_lora_dim * 2
) # Accommodate maximum stacking (gate_up_proj)
# # Allocate all MoE buffers with the same maximum rank dimension
# # to ensure consistent kernel compilation. The maximum stacking factor is 2.
# max_rank_dim = (
# max_lora_dim * 2
# ) # Accommodate maximum stacking (gate_up_proj)
# return (
# self.max_loras_per_batch,
# num_experts,
# max_rank_dim,
# input_dim,
# )
if self.is_moe_module(module_name):
c = get_stacked_multiply(module_name) # gate_up_proj_moe=2, down_proj_moe=1
return (
self.max_loras_per_batch,
num_experts,
max_rank_dim,
max_lora_dim * c, # according to the actual module to allocate
input_dim,
Comment on lines +172 to 189

Copilot AI Jan 28, 2026

Copy link

Choose a reason for hiding this comment

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

This change leaves a sizable commented-out alternative implementation/return block in get_lora_A_shape, and also includes a redundant nested if self.is_moe_module(module_name) inside an already-true MoE branch. Please remove the commented-out block and simplify the control flow (compute c = get_stacked_multiply(module_name) once) to keep the allocator logic clear.

Copilot uses AI. Check for mistakes.
)
else:
Expand Down
Loading
Loading