4 papers
SPAR: Support-Preserving Action Rectification
Jiaxin Zhao, Weihang Pan, Xun Liang +1
Offline policy improvement faces an inherent conflict between maximizing value and fitting the data distribution. While in-sample weighted regression is stable, it suffers from ove…
TokenSqueeze: Performance-Preserving Compression for Reasoning LLMs
Yuxiang Zhang, Zhengxu Yu, Weihang Pan +5
Emerging reasoning LLMs such as OpenAI-o1 and DeepSeek-R1 have achieved strong performance on complex reasoning tasks by generating long chain-of-thought (CoT) traces. However, the…
Enhancing Spatial Reasoning through Visual and Textual Thinking
Xun Liang, Xin Guo, Zhongming Jin +5
The spatial reasoning task aims to reason about the spatial relationships in 2D and 3D space, which is a fundamental capability for Visual Question Answering (VQA) and robotics. Al…
Beyond Templates: Dynamic Adaptation of Reasoning Demonstrations via Feasibility-Aware Exploration
Yong Wu, Weihang Pan, Ke Li +3
Large language models (LLMs) have shown remarkable reasoning capabilities, yet aligning such abilities to small language models (SLMs) remains a challenge due to distributional mis…