Showing cs.LGShow all
2 papers · 1 filter
cs.LG2026
Kalman Meets Curriculum: Efficient Dynamic Prompt Selection for Adaptive RL Finetuning
Haodong Zhu, Yangyang Ren, Yanjing Li +4
Reinforcement learning (RL) finetuning significantly enhances the reasoning capabilities of large language models (LLMs), yet its effectiveness critically depends on selecting prom…
cs.LG2026
Unbiased Dynamic Pruning for Efficient Group-Based Policy Optimization
Haodong Zhu, Yangyang Ren, Yanjing Li +6
Group Relative Policy Optimization (GRPO) effectively scales LLM reasoning but incurs prohibitive computational costs due to its extensive group-based sampling requirement. While r…