Jiachuan Wang

VoG: Enhancing LLM Reasoning through Stepwise Verification on Knowledge Graphs

Finding the Thread: Context-Driven Incremental Compression for Multi-Turn Dialogue

Activation-aware Probe-Query: Effective Key-Value Retrieval for Long-Context LLMs Inference

ReactionLLM: Data-Centric Learning for a Compact Single-step Chemical reaction prediction model

Structuring Benchmark into Knowledge Graphs to Assist Large Language Models in Retrieving and Designing Models

Learning Towards Emergence: Paving the Way to Induce Emergence by Inhibiting Monosemantic Neurons on Pre-trained Models