-
# What We Know About The Voynich Manuscript
-
- https://arxiv.org/abs/2108.11591
- EMNLP 2021
読み順検出は、視覚的に豊かな文書(領収書やフォームなど)を理解するための礎となる。
残念ながら、十分な規模のデータセットにアノテーションを施すのは手間がかかりすぎるため、高度な深層学習モデルを活用した既存の研究はありませんでした。
私たちは、WORD文書の読み順がそのXMLメタデータに埋…
e4exp updated
2 years ago
-
- https://arxiv.org/abs/2109.02377
- 2021 EMNLP
最近のTransformerのバリエーションであるPerformerは、Transformerを直線的な注目メカニズムでより長いシーケンスに対応させています。
しかし、Performerは、絶対位置エンコーディングよりも優れている相対位置エンコーディングに対応していません。
この論文では、…
e4exp updated
3 years ago
-
Migrated from google doc
-
## 0. Paper
@inproceedings{zhao-etal-2019-moverscore,
title = "{M}over{S}core: Text Generation Evaluating with Contextualized Embeddings and Earth Mover Distance",
author = "Zhao, Wei and
…
a1da4 updated
4 years ago
-
Hello,
I hope this message finds you well. Your work has been incredibly helpful to me. I wanted to inquire about the expected timeline for the release of the source code. Could you please provide …
-
- https://arxiv.org/abs/2108.12284
- 2021 EMNLP
近年、ニューラルネットワークの系統的な一般化能力をテストするために、多くのデータセットが提案されている。
その中で、標準的なタスクから得られるデフォルトのハイパーパラメータを用いて学習されたベースラインTransformerは、劇的に失敗することが示されている。
ここでは、埋め込みのスケーリ…
e4exp updated
3 years ago
-
- https://arxiv.org/abs/2009.11264
- EMNLP 2020
トランスフォーマーは、多くのNLPタスクでリカレントモデルに取って代わりました。
しかし、異なる構文特性をモデル化する能力の違いについては、まだほとんど知られていません。
過去の研究では、LSTMは通常の言語で非常によく一般化し、対言語との密接な関係があることが示唆されている。
本研究では…
e4exp updated
3 years ago
-
Struggling to get to the step of lecture yesterday at this point of the [recorded lecture](https://www.youtube.com/watch?v=JKvVgmG4ubE&list=PLSNWQVdrBwoa4KNaiKr-ayUdROZdSZ_1E&index=3). For some reaso…
-
## 0. Paper
@inproceedings{chen-etal-2014-unified,
title = "A Unified Model for Word Sense Representation and Disambiguation",
author = "Chen, Xinxiong and
Liu, Zhiyuan and
…
a1da4 updated
3 years ago