Jonathan Light

Strategist: Self-improvement of LLM Decision Making via Bi-Level Tree Search

2024 · arXiv (Cornell University)

Traditional reinforcement learning and planning typically requires vast amounts of data and training to develop effective policies. In contrast, large language models (LLMs) exhibit strong generalization and zero-shot capabilities, but struggle with tasks that require …

DISC: Dynamic Decomposition Improves LLM Inference Scaling

2025 · arXiv (Cornell University)

Inference scaling methods for LLMs often rely on decomposing problems into steps (or groups of tokens), followed by sampling and selecting the best next steps. However, these steps and their sizes are often predetermined or …