Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

add several recent embedding models #377

Merged
merged 1 commit into from
Oct 12, 2024
Merged

add several recent embedding models #377

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
54 changes: 37 additions & 17 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -265,24 +265,42 @@
| [local-politics-BERT](http://local-politics.jp/%e5%85%ac%e9%96%8b%e7%89%a9/local-politics-bert/) | 政治 | BERT (base) | Wikipedia, 国会会議録, 地方議会会議録 | 地方議会会議録コーパスプロジェクト | CC BY-SA 4.0 | ◯ ([SC-min](https://huggingface.co/local-politics-jp/bert-base-japanese-minutes-scratch), [SC-minwiki](https://huggingface.co/local-politics-jp/bert-base-japanese-minutes-wikipedia-scratch), [SC-2M-wiki](https://huggingface.co/local-politics-jp/bert-base-japanese-wikipedia-scratch-2m), [SC-2M-min](https://huggingface.co/local-politics-jp/bert-base-japanese-minutes-scratch-2m), [SC-2M-minwiki](https://huggingface.co/local-politics-jp/bert-base-japanese-minutes-wikipedia-scratch-2m), [FP-min](https://huggingface.co/local-politics-jp/bert-base-japanese-minutes-further), [FP-minwiki](https://huggingface.co/local-politics-jp/bert-base-japanese-minutes-wikipedia-further)) [^18] |

<a id="embeddings"></a>
## 埋め込み (Embeddings) 作成に特化したモデル
## 埋め込み (Embeddings) 作成に特化したモデル [^21]

| | アーキテクチャ | 開発元 | ライセンス |
### Bi-Encoders

#### Single-representation bi-encoders

| | 入力で扱えるトークン数 | 開発元 | ライセンス |
|:---|:---:|:---:|:---:|
| [JaColBERTv2.5](https://www.answer.ai/posts/2024-08-02-jacolbert-v25.html)<br>([JaColBERTv2.4](https://huggingface.co/answerdotai/JaColBERTv2.4), [JaColBERTv2.5](https://huggingface.co/answerdotai/JaColBERTv2.5)) | ColBERTv2 | Answer.AI | MIT |
| [JaColBERTv2](https://huggingface.co/bclavie/JaColBERTv2)<br>([JaColBERTv2](https://huggingface.co/bclavie/JaColBERTv2)) | ColBERTv2 | 個人 ([Benjamin Clavié](https://scholar.google.com/citations?user=vuMln98AAAAJ)) | MIT |
| [JaColBERT](https://arxiv.org/pdf/2312.16144.pdf)<br>([JaColBERT](https://huggingface.co/bclavie/JaColBERT)) | ColBERTv2 | 個人 ([Benjamin Clavié](https://scholar.google.com/citations?user=vuMln98AAAAJ)) | MIT |
| [Japanese SimCSE](https://github.com/hppRC/simple-simcse-ja)<br>([cl-nagoya/unsup-simcse-ja-base](https://huggingface.co/cl-nagoya/unsup-simcse-ja-base), [cl-nagoya/unsup-simcse-ja-large](https://huggingface.co/cl-nagoya/unsup-simcse-ja-large), [cl-nagoya/sup-simcse-ja-base](https://huggingface.co/cl-nagoya/sup-simcse-ja-base), [cl-nagoya/sup-simcse-ja-large](https://huggingface.co/cl-nagoya/sup-simcse-ja-large)) | SimCSE | 名大 武田・笹野研 | CC BY-SA 4.0 |
| [GLuCoSE](https://prtimes.jp/main/html/rd/p/000000123.000022705.html)<br>([pkshatech/GLuCoSE-base-ja](https://huggingface.co/pkshatech/GLuCoSE-base-ja)) | LUKEベースの文埋め込みモデル<br>(GLuCoSE) | PKSHA Technology | Apache 2.0 |
||||
| [colorfulscoop/sbert-base-ja](https://huggingface.co/colorfulscoop/sbert-base-ja) | Sentence-BERT | Colorful Scoop | CC BY-SA 4.0 |
| [MU-Kindai/SBERT-JSNLI-base](https://huggingface.co/MU-Kindai/SBERT-JSNLI-base)<br>[MU-Kindai/SBERT-JSNLI-large](https://huggingface.co/MU-Kindai/SBERT-JSNLI-large) | Sentence-BERT | 近畿大学 (研究室不明) | ？ |
| [MU-Kindai/Japanese-SimCSE-BERT-base-unsup](https://huggingface.co/MU-Kindai/Japanese-SimCSE-BERT-base-unsup)<br>[MU-Kindai/Japanese-SimCSE-BERT-large-unsup](https://huggingface.co/MU-Kindai/Japanese-SimCSE-BERT-large-unsup)<br>[MU-Kindai/Japanese-SimCSE-RoBERTa-base-unsup](https://huggingface.co/MU-Kindai/Japanese-SimCSE-RoBERTa-base-unsup)<br>[MU-Kindai/Japanese-SimCSE-BERT-base-sup](https://huggingface.co/MU-Kindai/Japanese-SimCSE-BERT-base-sup)<br>[MU-Kindai/Japanese-SimCSE-BERT-large-sup](https://huggingface.co/MU-Kindai/Japanese-SimCSE-BERT-large-sup) | SimCSE | 近畿大学 (研究室不明) | MIT |
| [pkshatech/simcse-ja-bert-base-clcmlp](https://huggingface.co/pkshatech/simcse-ja-bert-base-clcmlp) | SimCSE | PKSHA Technology | CC BY-SA 4.0 |
| [MU-Kindai/Japanese-MixCSE-BERT-base](https://huggingface.co/MU-Kindai/Japanese-MixCSE-BERT-base)<br>[MU-Kindai/Japanese-MixCSE-BERT-large](https://huggingface.co/MU-Kindai/Japanese-MixCSE-BERT-large) | MixCSE | 近畿大学 (研究室不明) | MIT |
| [MU-Kindai/Japanese-DiffCSE-BERT-base](https://huggingface.co/MU-Kindai/Japanese-DiffCSE-BERT-base) | DiffCSE | 近畿大学 (研究室不明) | MIT |
| [bclavie/fio-base-japanese-v0.1](https://huggingface.co/bclavie/fio-base-japanese-v0.1) | | 個人 ([Benjamin Clavié](https://scholar.google.com/citations?user=vuMln98AAAAJ)) | |
| [cl-nagoya/shioriha-large-pt](https://huggingface.co/cl-nagoya/shioriha-large-pt) | | 名大 武田・笹野研 | |
| [RoSEtta](https://prtimes.jp/main/html/rd/p/000000169.000022705.html)<br>([pkshatech/RoSEtta-base-ja](https://huggingface.co/pkshatech/RoSEtta-base-ja)) | 1,024 | PKSHA Technology | Apache 2.0 |
| [GLuCoSE v2](https://prtimes.jp/main/html/rd/p/000000169.000022705.html)<br>([pkshatech/GLuCoSE-base-ja-v2](https://huggingface.co/pkshatech/GLuCoSE-base-ja-v2)) | 512 | PKSHA Technology | Apache 2.0 |
| [Ruri](https://arxiv.org/abs/2409.07737)<br>([cl-nagoya/ruri-pt-small](https://huggingface.co/cl-nagoya/ruri-pt-small), [cl-nagoya/ruri-pt-base](https://huggingface.co/cl-nagoya/ruri-pt-base), [cl-nagoya/ruri-pt-large](https://huggingface.co/cl-nagoya/ruri-pt-large), [cl-nagoya/ruri-small](https://huggingface.co/cl-nagoya/ruri-small), [cl-nagoya/ruri-base](https://huggingface.co/cl-nagoya/ruri-base), [cl-nagoya/ruri-large](https://huggingface.co/cl-nagoya/ruri-large)) | 512 | 名大 笹野研 | Apache 2.0 |
| [Japanese SimCSE](https://github.com/hppRC/simple-simcse-ja)<br>([cl-nagoya/unsup-simcse-ja-base](https://huggingface.co/cl-nagoya/unsup-simcse-ja-base), [cl-nagoya/unsup-simcse-ja-large](https://huggingface.co/cl-nagoya/unsup-simcse-ja-large), [cl-nagoya/sup-simcse-ja-base](https://huggingface.co/cl-nagoya/sup-simcse-ja-base), [cl-nagoya/sup-simcse-ja-large](https://huggingface.co/cl-nagoya/sup-simcse-ja-large)) | 512 | 名大 笹野研 | CC BY-SA 4.0 |
| [GLuCoSE](https://prtimes.jp/main/html/rd/p/000000123.000022705.html)<br>([pkshatech/GLuCoSE-base-ja](https://huggingface.co/pkshatech/GLuCoSE-base-ja)) | 512 | PKSHA Technology | Apache 2.0 |
| [colorfulscoop/sbert-base-ja](https://huggingface.co/colorfulscoop/sbert-base-ja) || Colorful Scoop | CC BY-SA 4.0 |
| [MU-Kindai/SBERT-JSNLI-base](https://huggingface.co/MU-Kindai/SBERT-JSNLI-base)<br>[MU-Kindai/SBERT-JSNLI-large](https://huggingface.co/MU-Kindai/SBERT-JSNLI-large) || 近畿大学 (研究室不明) | ？ |
| [MU-Kindai/Japanese-SimCSE-BERT-base-unsup](https://huggingface.co/MU-Kindai/Japanese-SimCSE-BERT-base-unsup)<br>[MU-Kindai/Japanese-SimCSE-BERT-large-unsup](https://huggingface.co/MU-Kindai/Japanese-SimCSE-BERT-large-unsup)<br>[MU-Kindai/Japanese-SimCSE-RoBERTa-base-unsup](https://huggingface.co/MU-Kindai/Japanese-SimCSE-RoBERTa-base-unsup)<br>[MU-Kindai/Japanese-SimCSE-BERT-base-sup](https://huggingface.co/MU-Kindai/Japanese-SimCSE-BERT-base-sup)<br>[MU-Kindai/Japanese-SimCSE-BERT-large-sup](https://huggingface.co/MU-Kindai/Japanese-SimCSE-BERT-large-sup) || 近畿大学 (研究室不明) | MIT |
| [pkshatech/simcse-ja-bert-base-clcmlp](https://huggingface.co/pkshatech/simcse-ja-bert-base-clcmlp) || PKSHA Technology | CC BY-SA 4.0 |
| [MU-Kindai/Japanese-MixCSE-BERT-base](https://huggingface.co/MU-Kindai/Japanese-MixCSE-BERT-base)<br>[MU-Kindai/Japanese-MixCSE-BERT-large](https://huggingface.co/MU-Kindai/Japanese-MixCSE-BERT-large) || 近畿大学 (研究室不明) | MIT |
| [MU-Kindai/Japanese-DiffCSE-BERT-base](https://huggingface.co/MU-Kindai/Japanese-DiffCSE-BERT-base) || 近畿大学 (研究室不明) | MIT |
| [bclavie/fio-base-japanese-v0.1](https://huggingface.co/bclavie/fio-base-japanese-v0.1) || 個人 ([Benjamin Clavié](https://scholar.google.com/citations?user=vuMln98AAAAJ)) | |
| [cl-nagoya/shioriha-large-pt](https://huggingface.co/cl-nagoya/shioriha-large-pt) || 名大 笹野研 | |

#### Multi-representation bi-encoders

| | 開発元 | ライセンス |
|:---|:---:|:---:|
| [JaColBERTv2.5](https://www.answer.ai/posts/2024-08-02-jacolbert-v25.html)<br>([JaColBERTv2.4](https://huggingface.co/answerdotai/JaColBERTv2.4), [JaColBERTv2.5](https://huggingface.co/answerdotai/JaColBERTv2.5)) | Answer.AI | MIT |
| [JaColBERTv2](https://huggingface.co/bclavie/JaColBERTv2)<br>([JaColBERTv2](https://huggingface.co/bclavie/JaColBERTv2)) | 個人 ([Benjamin Clavié](https://scholar.google.com/citations?user=vuMln98AAAAJ)) | MIT |
| [JaColBERT](https://arxiv.org/pdf/2312.16144.pdf)<br>([JaColBERT](https://huggingface.co/bclavie/JaColBERT)) | 個人 ([Benjamin Clavié](https://scholar.google.com/citations?user=vuMln98AAAAJ)) | MIT |

### Cross-Encoders

| | 開発元 | ライセンス |
|:---|:---:|:---:|
| [Ruri-Reranker](https://arxiv.org/abs/2409.07737)<br>([cl-nagoya/ruri-reranker-stage1-small](https://huggingface.co/cl-nagoya/ruri-reranker-stage1-small), [cl-nagoya/ruri-reranker-stage1-base](https://huggingface.co/cl-nagoya/ruri-reranker-stage1-base), [cl-nagoya/ruri-reranker-stage1-large](https://huggingface.co/cl-nagoya/ruri-reranker-stage1-large), [cl-nagoya/ruri-reranker-small](https://huggingface.co/cl-nagoya/ruri-reranker-small), [cl-nagoya/ruri-reranker-base](https://huggingface.co/cl-nagoya/ruri-reranker-base), [cl-nagoya/ruri-reranker-large](https://huggingface.co/cl-nagoya/ruri-reranker-large)) | 名大 笹野研 | Apache 2.0 |
| [hotchpotch/japanese-reranker-cross-encoder-xsmall-v1](https://huggingface.co/hotchpotch/japanese-reranker-cross-encoder-xsmall-v1)<br>[hotchpotch/japanese-reranker-cross-encoder-small-v1](https://huggingface.co/hotchpotch/japanese-reranker-cross-encoder-small-v1)<br>[hotchpotch/japanese-reranker-cross-encoder-base-v1](https://huggingface.co/hotchpotch/japanese-reranker-cross-encoder-base-v1)<br>[hotchpotch/japanese-reranker-cross-encoder-large-v1](https://huggingface.co/hotchpotch/japanese-reranker-cross-encoder-large-v1)<br>[hotchpotch/japanese-bge-reranker-v2-m3-v1](https://huggingface.co/hotchpotch/japanese-bge-reranker-v2-m3-v1) | 個人 (舘野祐一) | MIT |

<a id="multimodal"></a>
## 視覚言語モデル (Vision-Language Models)
Expand Down Expand Up @@ -532,4 +550,6 @@

[^19]: 詳細は以下の記事を参照: [大規模言語モデルTanuki-8B, 8x8Bの位置づけや開発指針など](https://zenn.dev/matsuolab/articles/377f7ae8b1169e), [大規模言語モデルを開発するにあたっての事前・事後学習の戦略メモー特に合成データについてー](https://zenn.dev/matsuolab/articles/34036f017fae9e)

[^20]: ORPO を行う前に、Gemma 2 Instruct と Gemma 2 Base の差分の Chat Vector を加えている。
[^20]: ORPO を行う前に、Gemma 2 Instruct と Gemma 2 Base の差分の Chat Vector を加えている。

[^21]: 埋め込みモデルの分類は [Dense Text Retrieval based on Pretrained Language Models: A Survey (Zhao+, 2022)](https://arxiv.org/abs/2211.14876) を参考に行った。Bi-Encoder は 2つの入力を個別にモデルに入力し、それぞれベクトル化した上で、それらの内積やコサイン類似度を入力の近さとして定式化するアーキテクチャである。それに対し、Cross-Encoder は 2 つの入力を組み合わせたものをモデルに入力し、モデル内部で近さを直接計算するアーキテクチャである。情報抽出の分野では、Cross-Encoder の方が計算コストがかかるが、入力の近さをよりきめ細かくモデルが計算することが期待されるため、抽出結果の順序を再検討するリランカーとして用いられることも多い。なお、Bi-Encoder の中でも、入力を単一のベクトルではなく（トークンごとなどの）複数のベクトルとして表現するタイプのもの（例: ColBERT）があるため、Single-representation bi-encoders と Multi-representation bi-encoders にさらに細分化している。
Loading
Loading