DeepSeek R1 and DAPO

Overview

This presentation summarizes DeepSeek-R1 and DAPO, focusing on optimization objectives, training behavior, and practical implications for reasoning performance.

Papers

1. DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
2. DAPO: An Open-Source LLM Reinforcement Learning System at Scale
3. GDPO: Group reward-Decoupled Normalization Policy Optimization for Multi-reward RL Optimization

Slides

• Download PDF