arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2307.06962cs.CLcs.AI
收藏

复制就是你所需的一切

Copy Is All You Need

Tencent AI Lab(腾讯人工智能实验室);Beijing Institute of Technology(北京理工大学)

Tian Lan, Deng Cai, Yan Wang, Heyan Huang, Xian-Ling Mao

AI总结:

本文提出将文本生成转化为从文本集合中检索并复制文本片段的 Copy 框架,在 WikiText-103 上提升生成质量与效率,并支持免训练领域适应和规模扩展。

摘要:

主流文本生成模型通过从固定词表中依次选择词语来构成输出。本文将文本生成建模为从现有文本集合中逐步复制文本片段(例如词或短语)的过程。我们计算有意义文本片段的上下文化表示,并使用高效向量搜索工具包对其建立索引。随后,文本生成任务被分解为一系列复制粘贴操作:在每个时间步,我们从文本集合中寻找合适的文本跨度,而不是从独立词表中进行选择。在标准语言建模基准(WikiText-103)上的实验表明,根据自动评估和人工评估,我们的方法都取得了更好的生成质量。此外,由于解码步数减少,其推理效率可与 token 级自回归模型相媲美。我们还表明,只需切换到特定领域的文本集合而无需额外训练,我们的方法即可实现有效的领域适应。最后,我们观察到,同样无需进一步训练,只需扩展到更大的文本集合,我们的方法就能获得额外的性能提升。我们的源代码已公开发布于 https://github.com/gmftbyGMFTBY/Copyisallyouneed。

英文摘要:

The dominant text generation models compose the output by sequentially selecting words from a fixed vocabulary. In this paper, we formulate text generation as progressively copying text segments (e.g., words or phrases) from an existing text collection. We compute the contextualized representations of meaningful text segments and index them using efficient vector search toolkits. The task of text generation is then decomposed into a series of copy-and-paste operations: at each time step, we seek suitable text spans from the text collection rather than selecting from a standalone vocabulary. Experiments on the standard language modeling benchmark (WikiText-103) show that our approach achieves better generation quality according to both automatic and human evaluations. Besides, its inference efficiency is comparable to token-level autoregressive models thanks to the reduction of decoding steps. We also show that our approach allows for effective domain adaptation by simply switching to domain-specific text collection without extra training. Finally, we observe that our approach attains additional performance gains by simply scaling up to larger text collections, again without further training.\footnote{Our source codes are publicly available at \url{https://github.com/gmftbyGMFTBY/Copyisallyouneed}.}

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑