9 papers
Divergent-Convergent Reasoning: Scaling Test-Time Compute through Structured Solution Synthesis
Bo Wen, Yuhao Chen, Erhan Bilal +3
Test-time compute can substantially improve Large Language Model (LLM) reasoning performance, yet how and when additional compute helps remains poorly understood. We study Divergen…
STAIF: A Stage-wise Optimization for Complex Instruction Following
Jian Hong, Chen Cheng, Quan Liu +2
Following complex instructions with multiple explicit constraints remains a fundamental challenge for large language models (LLMs). Existing alignment methods, such as DPO, optimiz…
VehicleMemBench: An Executable Benchmark for Multi-User Long-Term Memory in In-Vehicle Agents
Yuhao Chen, Yi Xu, Xinyun Ding +7
With the growing demand for intelligent in-vehicle experiences, vehicle-based agents are evolving from simple assistants to long-term companions. This evolution requires agents to…
Efficiently Aligning Draft Models via Parameter- and Data-Efficient Adaptation
Luxi Lin, Zhihang Lin, Zhanpeng Zeng +5
Speculative decoding accelerates LLM inference but suffers from performance degradation when target models are fine-tuned for specific domains. A naive solution is to retrain draft…
PERMA: Benchmarking Personalized Memory Agents via Event-Driven Preference and Realistic Task Environments
Shuochen Liu, Junyi Zhu, Long Shu +11
Empowering large language models with long-term memory is crucial for building agents that adapt to users' evolving needs. Existing evaluations of this capability typically interle…
Look as You Think: Unifying Reasoning and Visual Evidence Attribution for Verifiable Document RAG via Reinforcement Learning
Shuochen Liu, Pengfei Luo, Chao Zhang +6
Aiming to identify precise evidence sources from visual documents, visual evidence attribution for visual document retrieval-augmented generation (VD-RAG) ensures reliable and veri…
TeaRAG: A Token-Efficient Agentic Retrieval-Augmented Generation Framework
Chao Zhang, Yuhao Wang, Derong Xu +9
Retrieval-Augmented Generation (RAG) utilizes external knowledge to augment Large Language Models' (LLMs) reliability. For flexibility, agentic RAG employs autonomous, multi-round…
Xiangqi-R1: Enhancing Spatial Strategic Reasoning in LLMs for Chinese Chess via Reinforcement Learning
Yuhao Chen, Shuochen Liu, Yuanjie Lyu +3
Game playing has long served as a fundamental benchmark for evaluating Artificial General Intelligence. While Large Language Models (LLMs) have demonstrated impressive capabilities…
Streamlining the Collaborative Chain of Models into A Single Forward Pass in Generation-Based Tasks
Yuanjie Lyu, Chao Zhang, Yuhao Chen +2
In Retrieval-Augmented Generation (RAG) and agent-based frameworks, the "Chain of Models" approach is widely used, where multiple specialized models work sequentially on distinct s…