AIPython★ 581
redai-studio/Relax
RELAX: ASYNCHRONOUS RL FOR MULTIMODAL POST‑TRAINING
Relax — an asynchronous reinforcement learning engine for scalable post‑training agents. It supports efficient distributed training, multi-agent scenarios, and working with various modalities via GRPO.
// KEY FEATURES
- Asynchronously trains agents using the GRPO to speed up the convergence.
- Provides scalable distributed training acrossnumerous nodes ensuring the fault tolerance.
- Allows the agents to collaborate in text, image, and audio modalities.
- Performs post‑training of the LLM models, preserving knowledge and adapting.
#agentic-rl#distributed-training#grpo#multi-agent#multimodal#post-training#ray-serve#reinforcement-learning#rlhf
Open on GitHub →New repositories every 30 minutes
REDDYX AI scans GitHub 24/7 and ships the best AI/ML/Web3 projects to Telegram.
Join on Telegram