# Large Language Models to Diffusion Finetuning

arXiv: https://arxiv.org/abs/2501.15781 · HTML: https://mlsanity.com/paper/2501.15781/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Placed in Diffusion reasoning and test-time scaling, one of 17 directions in LLM reasoning.
- First posted on arXiv 27 Jan 2025.

## Placed in

- LLM reasoning / [Diffusion reasoning and test-time scaling](https://mlsanity.com/lr/diffusion-reasoning-and-test-time-scaling/)

## Built on by (most cited first, 6 of 6)

- [A Survey on Latent Reasoning](https://mlsanity.com/paper/2507.06203/) (2025)
- [A Survey on Diffusion Language Models](https://mlsanity.com/paper/2508.10875/) (2025)
- [CANDI: Hybrid Discrete-Continuous Diffusion Models](https://mlsanity.com/paper/2510.22510/) (2025)
- [LaDiR: Latent Diffusion Enhances LLMs for Text Reasoning](https://mlsanity.com/paper/2510.04573/) (2025)
- [Watermarking Diffusion Language Models](https://mlsanity.com/paper/2509.24368/) (2025)
- [ELF-REG: Scaling Continuous Diffusion Language Models to Reasoning Tasks](https://mlsanity.com/paper/2609.29102/) (2026)

## Abstract (arXiv)

We propose a new finetuning method to provide pre-trained large language models (LMs) the ability to scale test-time compute through the diffusion framework. By increasing the number of diffusion steps, we show our finetuned models achieve monotonically increasing accuracy, directly translating to improved performance across downstream tasks. Furthermore, our finetuned models can expertly answer questions on specific topics by integrating powerful guidance techniques, and autonomously determine the compute required for a given problem by leveraging adaptive ODE solvers. Our method is universally applicable to any foundation model pre-trained with a cross-entropy loss and does not modify any of its original weights, fully preserving its strong single-step generation capabilities. We show our method is more effective and fully compatible with traditional finetuning approaches, introducing an orthogonal new direction to unify the strengths of the autoregressive and diffusion frameworks.