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 2010.07375cs.CL
收藏

神经叙事生成的解码方法

Decoding Methods for Neural Narrative Generation

Johns Hopkins University(约翰斯·霍普金斯大学);Stanford University(斯坦福大学);New York University(纽约大学)

Alexandra DeLucia, Aaron Mueller, Xiang Lisa Li, João Sedoc

摘要:

叙事生成是一个开放式的NLP任务,其中模型在给定提示的情况下生成故事。该任务类似于聊天机器人的神经反应生成;然而,响应生成的创新通常不适用于叙事生成,尽管这些任务之间存在相似性。我们的目标是通过应用和评估神经反应生成到神经叙事生成的解码方法的进展来弥合这一差距。特别是,我们采用GPT-2并在细胞核采样阈值和不同的解码超参数(特别是最大互信息)之间执行消融,通过自动和人工评估的多个标准分析结果。我们发现:(1)核采样通常是最好的阈值之间的0.7和0.9;(2)最大的互信息目标可以提高生成的故事的质量;(3)建立自动度量不相关以及与人类判断的叙事质量的任何定性指标。

英文摘要:

Narrative generation is an open-ended NLP task in which a model generates a story given a prompt. The task is similar to neural response generation for chatbots; however, innovations in response generation are often not applied to narrative generation, despite the similarity between these tasks. We aim to bridge this gap by applying and evaluating advances in decoding methods for neural response generation to neural narrative generation. In particular, we employ GPT-2 and perform ablations across nucleus sampling thresholds and diverse decoding hyperparameters -- specifically, maximum mutual information -- analyzing results over multiple criteria with automatic and human evaluation. We find that (1) nucleus sampling is generally best with thresholds between 0.7 and 0.9; (2) a maximum mutual information objective can improve the quality of generated stories; and (3) established automatic metrics do not correlate well with human judgments of narrative quality on any qualitative metric.

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。

↑