arXiv 2502.21074

CODI: Compressing Chain-of-Thought into Continuous Space via Self-Distillation

By Zhenyi Shen, Hanqi Yan, et al.

Published 2025-02-28

Wiki summary

Explore the paper's summary, context, and related research on Papiers.

Chain-of-Thought (CoT) reasoning enhances Large Language Models (LLMs) by encouraging step-by-step reasoning in natural language. However, leveraging a latent continuous space for reasoning may offer benefits in terms of both efficiency and robustness. Prior implicit CoT methods attempt to bypass language completely by reasoning in continuous space but have consistently underperformed compared to the standard explic…

View the original paper on arXiv