Zequn Sun

Small language models can now master complex tool use by learning from teachers without inheriting their biases or forgetting their own exploration gains.

Dense process supervision can transform RL training by providing step-level rewards that enhance credit assignment, leading to superior performance in complex reasoning tasks.

OaK transforms LLM agents by grounding their decision-making processes in dynamically constructed ontologies, leading to enhanced reliability in multi-step reasoning.

Explicitly leveraging structural context in entity alignment can significantly enhance performance, with ContextEA outperforming fine-tuned models on unseen knowledge graphs.