作者位次 #2/8ACL 2026Poster· Natural Language Processing
Distributional Clarity: The Hidden Driver of RL -Friendliness in Large Language Models
Shaoning Sun · Mingzhu Cai · Huang He · Bingjin Chen · Siqi Bao · Yujiu Yang · hua wu · Haifeng Wang
摘要
Language model families exhibit striking disparity in their capacity to benefit from reinforcement learning: under identical training, models like Qwen achieve substantial gains, while others like Llama yield limited improvements. Complementing data-centric approaches, we reveal that this dispari...