Magellan: Guided MCTS for Latent Space Exploration and Novelty Generation

TL;DR

Magellan introduces a guided exploration framework using MCTS and hierarchical guidance to enhance LLMs' ability to generate innovative and plausible scientific ideas, surpassing previous methods.

Contribution

The paper presents Magellan, a novel MCTS-based framework with hierarchical guidance for principled latent space exploration in LLMs, improving creativity and idea generation.

Findings

Magellan outperforms ReAct and ToT in scientific idea generation.

It achieves higher plausibility and innovation in generated ideas.

The framework effectively balances coherence, novelty, and narrative progress.

Abstract

Large Language Models (LLMs) often struggle with generating truly innovative ideas, typically defaulting to high-probability, familiar concepts within their training data's "gravity wells." While advanced search-based methods like Tree of Thoughts (ToT) attempt to mitigate this, they are fundamentally limited by their reliance on unprincipled, inconsistent self-evaluation heuristics to guide exploration. To address this gap, we introduce \textbf{Magellan}, a novel framework that reframes creative generation as a principled, guided exploration of an LLM's latent conceptual space. At its core, Magellan employs Monte Carlo Tree Search (MCTS) governed by a hierarchical guidance system. For long-range direction, a "semantic compass" vector, formulated via orthogonal projection, steers the search towards relevant novelty. For local, step-by-step decisions, a landscape-aware value function replaces flawed self-evaluation with an explicit reward structure that balances intrinsic coherence, extrinsic novelty, and narrative progress. Extensive experiments demonstrate that Magellan significantly outperforms strong baselines, including ReAct and ToT, in generating scientific ideas with superior plausibility and innovation. Our work shows that for creative discovery, a principled, guided search is more effective than unconstrained agency, paving the way for LLMs to become more capable partners in innovation.