高效一对多翻译：联合多流扩散
Efficient One-to-Many Translation with Joint Multi-Stream Diffusion
- Worcester Polytechnic Institute（伍斯特理工学院）
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
针对一对多翻译中自回归系统延迟高的问题，提出联合多流离散扩散框架，并行优化所有目标语言，实现亚线性延迟、统一模型部署及零样本迁移，在加速采样下达到AR质量并获2倍加速和零样本BLEU提升11.9%。
AI中文摘要：
一对多机器翻译（MT）对于自回归（AR）系统而言计算成本高昂，因为其延迟随序列长度和目标语言数量线性扩展。我们探索了扩散如何通过一种离散扩散框架实现多语言翻译，该框架并行优化所有目标语言，实现延迟随目标数量亚线性扩展，并支持作为单一统一模型部署以替代多个独立系统。该框架以连续语义锚点而非源词元为条件，支持无需重新训练即可零样本迁移到未见过的源语言，在零样本源上保持其监督翻译质量的约75%。我们研究了质量-延迟前沿，发现通过加速采样，其监督质量可与AR基线相当，同时获得2倍加速，并在零样本BLEU上提升11.9%。这些结果凸显了联合多流扩散作为高效一对多翻译的实用且灵活替代方案的潜力。
英文摘要：
One-to-many machine translation (MT) is computationally expensive for autoregressive (AR) systems, which suffer from linear latency scaling with both sequence length and the number of target languages. We explore how diffusion can enable multilingual translation with a discrete diffusion framework that refines all target languages in parallel, achieving sublinear latency scaling with the number of targets, and supports deployment as a single unified model to replace multiple independent systems. Conditioned on a continuous semantic anchor rather than source tokens, our framework supports zero-shot transfer to unseen source languages without retraining, maintaining approximately $75\%$ of its supervised translation quality on zero-shot sources. We investigate the quality-latency frontier and find that with accelerated sampling, it achieves comparable supervised quality to AR baselines with a $2 \times$ speedup and $11.9\%$ better zero-shot BLEU. These results highlight the potential of joint multi-stream diffusion as a practical and flexible alternative for efficient one-to-many translation.
