【文件属性】:
文件名称:Deep Learning Approaches to Text Production
文件大小:9.81MB
文件格式:PDF
更新时间:2022-05-08 11:17:47
Text
Uses the open source tf-seq2seq framework to train a char2char model on
the E2E NLG Challenge data.
No delexicalization, lowercasing or even tokenization
Input semantics = sequence of characters
Human evaluation shows that
● The output is grammatically perfect
● The model does not generate non words