Unlocking the Potential of Diffusion Language Models through Template Infilling

ACL ARR 2026 January Submission3981 Authors

Keywords: Diffusion language model

Abstract: Diffusion Language Models (DLMs) have emerged as a promising alternative to Autoregressive Language Models, yet their inference strategies remain limited to prefix-based prompting inherited from the autoregressive paradigm. In this paper, we propose Template Infilling (TI), a tailored conditioning methodology for DLMs. Unlike conventional prefix prompting, TI flexibly aligns structural anchors across the entire target response space, establishing a global blueprint before filling in the masked segments. We demonstrate the effectiveness of our approach on diverse benchmarks, including mathematical reasoning, code generation, and trip planning, achieving consistent improvements of 9.40%p over the baseline. Furthermore, we observe that TI provides additional advantages in multi-token generation settings, enabling effective speedup while maintaining generation quality and robustness. By enforcing these global constraints, TI ultimately facilitates System-2 reasoning, empowering the model to deliberate within a structurally defined solution space.

Paper Type: Long

Research Area: Language Models

Research Area Keywords: Diffusion Language Models, Template Infilling, System 2 thinking, Constrained Generation

Contribution Types: Model analysis & interpretability, NLP engineering experiment

Languages Studied: English

Submission Number: 3981