提出卷积解码与拒绝式微调，让扩散语言模型更快更流畅。
Fast and Fluent Diffusion Language Models via Convolutional Decoding and Rejective Fine-tuning
- 用卷积方法动态缩窄解码窗口，避免分块带来的延迟
- 在AlpacaEval上超越现有扩散模型，步数减少一半以上
- 适合追求高速生成且注重文本连贯性的应用
自回归语言模型逐词生成，限制了推理速度。基于扩散的语言模型可并行生成多个词，是潜在替代方案。然而我们发现当前扩散语言模型存在长解码窗问题：远离输入上下文的词常变得无关或重复。此前的半自回归方法通过分块处理缓解此问题，但牺牲了双向性，并引发时间间隔扩展问题，反而削弱了速度优势。为此，我们提出卷积解码（Conv），一种基于归一化的窗口压缩方法，无需硬性分割，提升流畅性与灵活性。同时引入拒绝式规则微调（R2FT），一种后训练策略，优化远距离位置的词对齐。所提方法在开放式生成基准（如AlpacaEval）上达到扩散模型基线最优表现，且所需步数显著低于以往工作，验证了速度与质量的双重提升。
原文摘要 · Abstract (English)
Autoregressive (AR) language models generate text one token at a time, which limits their inference speed. Diffusion-based language models offer a promising alternative, as they can decode multiple tokens in parallel. However, we identify a key bottleneck in current diffusion LMs: the long decoding-window problem, where tokens generated far from the input context often become irrelevant or repetitive. Previous solutions like semi-autoregressive address this issue by splitting windows into blocks (sacrificing bidirectionality), but we find that this also leads to time-interval expansion problem, sacrificing the speed. Therefore, semi-AR eliminates the main advantages of diffusion models. To overcome this, we propose Convolutional decoding (Conv), a normalization-based method that narrows the decoding window without hard segmentation, leading to better fluency and flexibility. Additionally, we introduce Rejecting Rule-based Fine-Tuning (R2FT), a post-hoc training scheme that better aligns tokens at positions far from context. Our methods achieve state-of-the-art results on open-ended generation benchmarks (e.g., AlpacaEval) among diffusion LM baselines, with significantly lower step size than previous works, demonstrating both speed and quality improvements.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。