BandPO: Bridging Trust Regions and Ratio Clipping via Probability-Aware Bounds for LLM Reinforcement Learning
Paper
• 2603.04918 • Published
• 41
Zhiyuan Li
HaharryrY
·
AI & ML interests
None yet
Recent Activity
upvoted a paper about 1 month ago
HERMES: KV Cache as Hierarchical Memory for Efficient Streaming Video Understanding upvoted a paper about 2 months ago
ABC-Bench: Benchmarking Agentic Backend Coding in Real-World Development Organizations
