13 papers
Reasoning Can Be Restored by Correcting a Few Decision Tokens
Changshuo Shen, Leheng Sheng, Yuxin Chen +2
Large reasoning models (LRMs) substantially outperform their base LLM counterparts on challenging reasoning benchmarks, yet it remains poorly understood where base models go wrong…
Internalizing Safety Understanding in Large Reasoning Models via Verification
Yi Zhang, Yuxin Chen, Leheng Sheng +4
While explicit Chain-of-Thought (CoT) empowers large reasoning models (LRMs), it enables the generation of riskier final answers. Current alignment paradigms primarily rely on exte…
Fine-grained Semantics Integration for Large Language Model-based Recommendation
Jiawei Feng, Xiaoyu Kong, Leheng Sheng +8
Recent advances in Large Language Models (LLMs) have driven a shift in recommender systems from the discriminative paradigm to the LLM-based generative paradigm, where the recommen…
Reinforcing Chain-of-Thought Reasoning with Self-Evolving Rubrics
Leheng Sheng, Wenchang Ma, Ruixin Hong +3
Despite chain-of-thought (CoT) playing crucial roles in LLM reasoning, directly rewarding it is difficult: training a reward model demands heavy human labeling efforts, and static…
When to Memorize and When to Stop: Gated Recurrent Memory for Long-Context Reasoning
Leheng Sheng, Yongtao Zhang, Wenchang Ma +6
While reasoning over long context is crucial for various real-world applications, it remains challenging for large language models (LLMs) as they suffer from performance degradatio…
On Reasoning Strength Planning in Large Reasoning Models
Leheng Sheng, An Zhang, Zijian Wu +5
Recent studies empirically reveal that large reasoning models (LRMs) can automatically allocate more reasoning strengths (i.e., the number of reasoning tokens) for harder problems,…