Associate Professor, Leiden University
3 papers at NeurIPS 2025
We "clone" large LLMs into small SLMs by training only low-rank projection matrices for weights and making all student activations identical to the teacher's. This yields comparable SLM performance with 1000x fewer training tokens.
We propose a Belief-Calibrated Consensus Seeking (BCCS) framework to facilitate stable consensus in multi-agent system via selecting optimal collaborators and calibrating the consensus judgment by system-internal beliefs.
We propose ExSearch, an agentic search framework, where the LLM learns to retrieve useful information as the reasoning unfolds through a self-incentivized process.