Latent Chain-of-Thought as Planning: Decoupling Reasoning from Verbalization

TL;DR

This paper introduces PLaT, a latent reasoning framework that models reasoning as planning in continuous states, enabling dynamic termination and broader solution exploration, thereby improving scalability and transparency in complex problem solving.

Contribution

PLaT decouples reasoning from verbalization by modeling latent reasoning as planning, allowing dynamic inference termination and enhanced reasoning diversity.

Findings

PLaT achieves broader solution space and better scalability.

PLaT demonstrates transparent and flexible reasoning trajectories.

Lower greedy accuracy but improved reasoning diversity.

Abstract

Chain-of-Thought (CoT) empowers Large Language Models (LLMs) to tackle complex problems, but remains constrained by the computational cost and reasoning path collapse when grounded in discrete token spaces. Recent latent reasoning approaches attempt to optimize efficiency by performing reasoning within continuous hidden states. However, these methods typically operate as opaque end-to-end mappings from explicit reasoning steps to latent states, and often require a pre-defined number of latent steps during inference. In this work, we introduce PLaT (Planning with Latent Thoughts), a framework that reformulates latent reasoning as planning by fundamentally decouple reasoning from verbalization. We model reasoning as a deterministic trajectory of latent planning states, while a separate Decoder grounds these thoughts into text when necessary. This decoupling allows the model to dynamically determine when to terminate reasoning rather than relying on fixed hyperparameters. Empirical results on mathematical benchmarks reveal a distinct trade-off: while PLaT achieves lower greedy accuracy than baselines, it demonstrates superior scalability in terms of reasoning diversity. This indicates that PLaT learns a robust, broader solution space, offering a transparent and scalable foundation for inference-time search. Our code can be found in https://github.com/yunsaijc/PLaT.