dVoting让扩散语言模型无需训练就能快速提升推理能力。
dVoting: Fast Voting for dLLMs
- 通过多轮采样与一致性分析，动态识别并重生成不确定词元。
- 在GSM8K等4个基准上提升6.22%至14.84%，效果稳定显著。
- 适合追求高效推理增强的AI研究者与应用开发者。
扩散大语言模型（dLLMs）代表了超越自回归建模的新范式，能在任意位置并行生成词元，显著提升推理效率。本文提出dVoting，一种无需训练的快速投票技术，在可接受的计算开销下增强推理能力。该方法基于多个样本对同一提示的预测高度一致，仅少数词元存在跨样本差异。利用dLLMs的任意位置生成能力，dVoting通过迭代采样、一致性分析识别不确定词元、投票重生成并重复直至收敛。大量实验表明，dVoting在多个基准上持续提升性能：在GSM8K上提升6.22%-7.66%，在MATH500上提升4.40%-7.20%，在ARC-C上提升3.16%-14.84%，在MMLU上提升4.83%-5.74%。代码已开源。
原文摘要 · Abstract (English)
Diffusion Large Language Models (dLLMs) represent a new paradigm beyond autoregressive modeling, offering competitive performance while naturally enabling a flexible decoding process. Specifically, dLLMs can generate tokens at arbitrary positions in parallel, endowing them with significant potential for parallel test-time scaling, which was previously constrained by severe inefficiency in autoregressive modeling. In this work, we introduce dVoting, a fast voting technique that boosts reasoning capability without training, with only an acceptable extra computational overhead. dVoting is motivated by the observation that, across multiple samples for the same prompt, token predictions remain largely consistent, whereas performance is determined by a small subset of tokens exhibiting cross-sample variability. Leveraging the arbitrary-position generation capability of dLLMs, dVoting performs iterative refinement by sampling, identifying uncertain tokens via consistency analysis, regenerating them through voting, and repeating this process until convergence. Extensive evaluations demonstrate that dVoting consistently improves performance across various benchmarks. It achieves gains of 6.22%-7.66% on GSM8K, 4.40%-7.20% on MATH500, 3.16%-14.84% on ARC-C, and 4.83%-5.74% on MMLU. Our code is available at https://github.com/fscdc/dVoting
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。