opendilab / LightRFT

LightRFT: Light, Efficient, Omni-modal & Reward-model Driven Reinforcement Fine-Tuning Framework

Why this job: says: fine tuning; topics: llm-training, reinforcement-learning

What it does

A common interface connects DeepSpeed/FSDP v2 training backends with SGLang/vLLM rollout backends, reducing backend-specific coupling in the upper-level training workflow.
Distributed jobs follow a single-program, multiple-data (SPMD) topology launched with torchrun; Ray is not required for scheduling, and standard PyTorch distributed tools remain applicable for debugging.
Training and rollout reuse the same GPU process set by phase. The rollout engine can sleep during policy updates and receives refreshed Actor weights before the next rollout.
FSDP v2 and DeepSpeed ZeRO stages 1/2/3; DeepSpeed is selected when --fsdp is absent.
BF16, gradient checkpointing, Adam offload, and FSDP CPU offload.

git clone https://github.com/opendilab/LightRFT.git
cd LightRFT
pip install -e .

Repos similar to this one
Repo	Stars	Last commit	Licence	Downloads for
Human-Agent-Society / reef	7,794	today	Apache-2.0	No downloads
oumi-ai / oumi	9,430	yesterday	Apache-2.0	No downloads
chi2liu / ABC-GRPO	438	2 months ago	Apache-2.0	No downloads

More: languages, topics and project record