DRA-GRPO: Exploring Diversity-Aware Reward Adjustment for R1-Zero-Like Training of Large Language Models

GitHub

DRA-GRPO

HTTPS

SSH

git@github.com:xiwenc1/DRA-GRPO.git

CLI

gh repo clone xiwenc1/DRA-GRPO

Transform this paper into an audio lecture

Get an engaging lecture and Q&A format to quickly understand the paper in minutes, perfect for learning on the go.

Audio lecture

Q&A format

alphaXiv