-
### Summary of other models
| Model | Supported? | Paper | Source code | Vocals (SDR) | Drums (SDR) | Bass (SDR) | Other (SDR) | Avg (SDR) | Notes |
| -- | -- | -- | -- | -- | -- | -- | -- | -- | …
-
## Why
Machine Learning 輪講は最新の技術や論文を追うことで、エンジニアが「技術で解決できること」のレベルをあげていくことを目的にした会です。
prev. #87
## What
話したいことがある人はここにコメントしましょう!
面白いものを見つけた時点でとりあえず話すという宣言だけでもしましょう!
-
## Explaining Data Patterns in Natural Language with Language Models
2023 BlackboxNLP Workshop at ACL | MSR & Cornell U
不断的生成解释,并进行排序。找出一个最具解释性的prompt。
Explanation: symbolic regression,
## Auto…
-
## 0. 論文
| タイトル | DeBERTa: Decoding-enhanced BERT with Disentangled Attention |
|:-------|:--|
| リンク | https://arxiv.org/abs/2006.03654 , https://github.com/microsoft/DeBERTa |
| 著者 | Pengc…
-
- https://openreview.net/forum?id=onxoVA9FxMw
- 2021 ICLR
概要:
Transformer based architectures(BERTなど)において、単語の順序をモデル化するために、様々なPosition Embeddings(PE)が提案されています。
これらは経験に基づいており、良い性能を発揮しますが、それらを体系…
e4exp updated
2 years ago
-
- https://arxiv.org/abs/2105.14602
- 2021 ICLR
大規模なニューラルネットワークが学習データの記憶をどのように回避しているかを理解することは、その高い汎化性能を説明する鍵となる。
本研究では、ディープネットワークにおいて、いつ、どこで記憶が行われるかという構造を調べるために、最近開発されたレプリカベースの平均場理論的な幾何学的解析手法を用いる。…
e4exp updated
3 years ago
-
- https://arxiv.org/abs/2006.15595
- 2021 ICLR
本研究では、言語の事前学習(例:BERT)で使用される位置符号化法を調査し、既存の定式化におけるいくつかの問題点を明らかにする。
まず、絶対位置エンコーディングにおいて、位置エンベッディングと単語エンベッディングに適用される加算演算は、2つの異種の情報資源の間に混在した相関をもたらすことを示す…
e4exp updated
3 years ago
-
# ChatGPT is fun, but it is not funny! Humor is still challenging Large Language Models
2023 Workshop on Computational Approaches to Subjectivity, Sentiment
“oxymoron” Despite being fun to interact …
-
- [link to OpenReview](https://openreview.net/forum?id=U1edbV4kNu_)
- 2021 ICLR 2022 Conference Blind Submission
概要:
多くの深層学習アプリケーションでは、何十億ものパラメータを持つ大規模なモデルを使用することで利益を得ています。
これらのモデルは、低レイテンシーかつ高…
e4exp updated
2 years ago
-
- [ ] [cleanlab/README.md at master · cleanlab/cleanlab](https://github.com/cleanlab/cleanlab/blob/master/README.md?plain=1)
# cleanlab/README.md at master · cleanlab/cleanlab
cleanlab helps y…