Twist Decoding: Diverse Generators Guide Each Other
Twist Decoding: Diverse Generators Guide Each Other
复制标题
DOI：
10.48550/arxiv.2205.09273
复制
发表时间：
2022-05
影响因子：
2.4
通讯作者：
Jungo Kasai;Keisuke Sakaguchi;Ronan Le Bras;Hao Peng;Ximing Lu;Dragomir R. Radev;Yejin Choi;Noah A. Smith
中科院分区：
文献类型：
--
作者：
Jungo Kasai;Keisuke Sakaguchi;Ronan Le Bras;Hao Peng;Ximing Lu;Dragomir R. Radev;Yejin Choi;Noah A. Smith
许多语言生成模型现在可用于广泛的生成任务，包括机器翻译和摘要。组合这些不同的模型可能会带来进一步的进展，但集成生成模型在推理过程中具有挑战性：传统的集成方法（例如浅层融合）要求模型共享词汇/标记化方案。我们引入了 Twist 解码，这是一种简单且通用的文本生成算法，它在推理时受益于多种模型。我们的方法不假设词汇、标记化甚至生成顺序是共享的。我们对机器翻译和科学论文摘要的广泛评估表明，Twist 解码在各种场景下的性能明显优于单独解码的每个模型，包括特定领域模型和通用模型都可用的情况。 Twist 解码也始终优于流行的重新排序启发式算法，其中一个模型的输出候选由另一个模型重新评分。我们希望我们的工作将鼓励研究人员和从业者集体而不是独立地研究生成模型，并寻找与当前可用模型具有互补优势的模型。
Many language generation models are now available for a wide range of generation tasks, including machine translation and summarization. Combining such diverse models may lead to further progress, but ensembling generation models is challenging during inference: conventional ensembling methods (e.g., shallow fusion) require that the models share vocabulary/tokenization schemes. We introduce Twist decoding, a simple and general text generation algorithm that benefits from diverse models at inference time. Our method does not assume the vocabulary, tokenization or even generation order is shared. Our extensive evaluations on machine translation and scientific paper summarization demonstrate that Twist decoding substantially outperforms each model decoded in isolation over various scenarios, including cases where domain-specific and general-purpose models are both available. Twist decoding also consistently outperforms the popular reranking heuristic where output candidates from one model are rescored by another. We hope that our work will encourage researchers and practitioners to examine generation models collectively, not just independently, and to seek out models with complementary strengths to the currently available models.