[2503.14476] DAPO: An Open-Source LLM Reinforcement Learning System at Scale - discu.eu

Reddit

"DAPO: An Open-Source LLM Reinforcement Learning System at Scale", Yu et al. 2025 https://arxiv.org/abs/2503.14476 2 comments 25/3/2025 reinforcementlearning

Linking pages

Recent reasoning research: GRPO tweaks, base model RL, and data curation https://www.interconnects.ai/p/papers-im-reading-base-model-rl-grpo 0 comments

Related searches:

Search whole site: site:arxiv.org

Search title: [2503.14476] DAPO: An Open-Source LLM Reinforcement Learning System at Scale

See how to search.

Submit link to: