SoftCoT: Soft Chain-of-Thought for Efficient Reasoning with LLMs

TL;DR

SoftCoT introduces a novel method for continuous-space reasoning in LLMs using a fixed assistant model and a trainable projection, improving reasoning performance without modifying the original LLM.

Contribution

The paper proposes a lightweight, non-intrusive approach for soft chain-of-thought reasoning that enhances LLM performance via supervised, parameter-efficient fine-tuning.

Findings

Improved reasoning accuracy on five benchmarks.

Effective use of a fixed assistant model for soft thought generation.

Parameter-efficient fine-tuning enhances LLM reasoning capabilities.

Abstract

Chain-of-Thought (CoT) reasoning enables Large Language Models (LLMs) to solve complex reasoning tasks by generating intermediate reasoning steps. However, most existing approaches focus on hard token decoding, which constrains reasoning within the discrete vocabulary space and may not always be optimal. While recent efforts explore continuous-space reasoning, they often require full-model fine-tuning and suffer from catastrophic forgetting, limiting their applicability to state-of-the-art LLMs that already perform well in zero-shot settings with a proper instruction. To address this challenge, we propose a novel approach for continuous-space reasoning that does not require modifying the LLM. Specifically, we employ a lightweight fixed assistant model to speculatively generate instance-specific soft thought tokens as the initial chain of thoughts, which are then mapped into the LLM's representation space via a trainable projection module. Experimental results on five reasoning benchmarks demonstrate that our method enhances LLM reasoning performance through supervised, parameter-efficient fine-tuning. Source code is available at https://github.com/xuyige/SoftCoT.