Research
LaDiR: Latent Diffusion Enhances LLMs for Text Reasoning
Large Language Models (LLMs) demonstrate their reasoning ability through chain-of-thought (CoT) generation. However, LLM’s autoregressive decoding may limit the ability to revisit and refine earlier t
Large Language Models (LLMs) demonstrate their reasoning ability through chain-of-thought (CoT) generation. However, LLM’s autoregressive decoding may limit the ability to revisit and refine earlier tokens in a holistic manner, which can also lead to inefficient exploration for diverse solutions. In this paper, we propose LaDiR (Latent Diffusion Reasoner), a novel reasoning framework that unifies the expressiveness of continuous latent representation with the iterative refinement capabilities of latent diffusion models for an existing LLM. We first construct a structured latent reasoning space…
Related
- Rectifying LLM Thought from Lens of Optimization
- Stability-Weighted Decoding for Diffusion Language Models
- SeLaR: Selective Latent Reasoning in Large Language Models
- LEPO: nderline{L}atent Rnderline{e}asoning nderline{P}olicy nderline{O}ptimization for Large Language~Models
- On the Reasoning Abilities of Masked Diffusion Language Models
- Diffusion Language Models Know the Answer Before Decoding
Source: Apple ML Research | 2026-04-28