13 papers
Why Multi-Step Tool-Use Reinforcement Learning Collapses and How Supervisory Signals Fix It
Yupu Hao, Zhuoran Jin, Huanxuan Liao +2
Tool use enables large language models (LLMs) to perform complex tasks, and recent agentic reinforcement learning (RL) methods show promise for enhancing model capabilities. Howeve…
Pushing the Limits of LLM Tool Calling via Experiential Knowledge Integration and Activation
Yupu Hao, Zhuoran Jin, Huanxuan Liao +2
Large language models (LLMs) rely on tool use to act as autonomous agents, yet often fail in multi-step execution due to insufficient tool-related knowledge and ineffective knowled…
Bottom-up Policy Optimization: Your Language Model Policy Secretly Contains Internal Policies
Yuqiao Tan, Minzheng Wang, Shizhu He +6
Existing reinforcement learning (RL) approaches treat large language models (LLMs) as a unified policy, overlooking their internal mechanisms. In this paper, we decompose the LLM-b…
ResAdapt: Adaptive Resolution for Efficient Multimodal Reasoning
Huanxuan Liao, Zhongtao Jiang, Yupu Hao +6
Multimodal Large Language Models (MLLMs) achieve stronger visual understanding by scaling input fidelity, yet the resulting visual token growth makes jointly sustaining high spatia…
A Comprehensive Survey on Long Context Language Modeling
Jiaheng Liu, Dawei Zhu, Zhiqi Bai +34
Efficient processing of long contexts has been a persistent pursuit in Natural Language Processing. With the growing number of long documents, dialogues, and other textual data, it…
SparK: Query-Aware Unstructured Sparsity with Recoverable KV Cache Channel Pruning
Huanxuan Liao, Yixing Xu, Shizhu He +6
Long-context inference in large language models (LLMs) is increasingly constrained by the KV cache bottleneck: memory usage grows linearly with sequence length, while attention com…