Skip to content

Add ZeRO-3 leaf module support for Qwen MoE models #701

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
wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Add ZeRO-3 leaf module support for Qwen MoE models #701

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
10 changes: 10 additions & 0 deletions src/open_r1/sft.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -91,6 +91,16 @@ def main(script_args, training_args, model_args):
tokenizer = get_tokenizer(model_args, training_args)
model = get_model(model_args, training_args)

# Qwen MoE: set ZeRO-3 leaf modules (lazy import; no effect for non-Qwen)
if getattr(getattr(model, "config", {}), "model_type", "") == "qwen3_moe":
try:
from transformers.models.qwen3_moe.modeling_qwen3_moe import Qwen3MoeSparseMoeBlock as _QwenSparseMoeBlock
import deepspeed
deepspeed.utils.set_z3_leaf_modules(model, [_QwenSparseMoeBlock])
logger.info("[MoE] ZeRO-3 leaf module setup for Qwen MoE model completed.")
except Exception as e:
logger.warning(f"[MoE] Skipped ZeRO-3 leaf module setup: {e}")

if tokenizer.chat_template is None:
logger.info("No chat template provided, defaulting to ChatML.")
model, tokenizer = setup_chat_format(model, tokenizer, format="chatml")
Expand Down