Search papers, labs, and topics across Lattice.
This paper introduces a novel approach to length control in diffusion language models (DLMs) by framing the problem as a discrete-time survival analysis of the end-of-sequence token. The proposed training-free length predictor significantly enhances inference efficiency, achieving up to a 7-fold speedup while maintaining accuracy across various reasoning and code-generation tasks. Additionally, the study reveals that the optimal sequence length can vary considerably within the same dataset, highlighting the importance of adaptive length selection for model performance.
Survival-guided length decoding can accelerate DLM inference by up to 7 times without sacrificing accuracy, revealing the critical role of adaptive length selection.
Diffusion language models (DLMs) generate text by iteratively denoising masked sequences, but standard decoding either fixes the sequence length or relies on ad hoc stopping rules, often leading to unnecessary denoising steps. We recast length selection as a discrete-time survival problem over the end-of-sequence token and propose a plug-in, training-free length predictor that can be added to any existing DLM. Across reasoning and code-generation benchmarks, survival-guided length decoding speeds up inference by up to 7 times while preserving task accuracy. We further find that predicted lengths vary widely even within the same dataset, making model performance sensitive to the chosen length.
