Improved Language Modeling by Decoding the Past
收藏DataCite Commons2026-01-07 更新2025-04-16 收录
下载链接：
https://service.tib.eu/ldmservice/dataset/70afd526-120a-423c-bcec-20b2f69840d2
下载链接
链接失效反馈官方服务：
资源简介：
Highly regularized LSTMs achieve impressive results on several benchmark datasets in language modeling. We propose a new regularization method based on decoding the last token in the context using the predicted distribution of the next token.
提供机构：
TIB
创建时间：
2024-12-16
