| --- |
| license: mit |
| datasets: |
| - hotchpotch/JQaRA |
| - shunk031/JGLUE |
| - miracl/miracl |
| - castorini/mr-tydi |
| - unicamp-dl/mmarco |
| language: |
| - ja |
| library_name: sentence-transformers |
| pipeline_tag: text-ranking |
| new_version: hotchpotch/japanese-reranker-base-v2 |
| --- |
| |
| ## hotchpotch/japanese-reranker-cross-encoder-large-v1 |
|
|
| 日本語で学習させた Reranker (CrossEncoder) シリーズです。 |
|
|
| | モデル名 | layers | hidden_size | |
| | ----------------------------------------------------------------------------------------------------------------------------------- | ------ | ----------- | |
| | [hotchpotch/japanese-reranker-cross-encoder-xsmall-v1](https://huggingface.co/hotchpotch/japanese-reranker-cross-encoder-xsmall-v1) | 6 | 384 | |
| | [hotchpotch/japanese-reranker-cross-encoder-small-v1](https://huggingface.co/hotchpotch/japanese-reranker-cross-encoder-small-v1) | 12 | 384 | |
| | [hotchpotch/japanese-reranker-cross-encoder-base-v1](https://huggingface.co/hotchpotch/japanese-reranker-cross-encoder-base-v1) | 12 | 768 | |
| | [hotchpotch/japanese-reranker-cross-encoder-large-v1](https://huggingface.co/hotchpotch/japanese-reranker-cross-encoder-large-v1) | 24 | 1024 | |
| | [hotchpotch/japanese-bge-reranker-v2-m3-v1](https://huggingface.co/hotchpotch/japanese-bge-reranker-v2-m3-v1) | 24 | 1024 | |
| |
| Reranker についてや、技術レポート・評価等は以下を参考ください。 |
| |
| - [日本語最高性能のRerankerをリリース / そもそも Reranker とは?](https://secon.dev/entry/2024/04/02/070000-japanese-reranker-release/) |
| - [日本語 Reranker 作成のテクニカルレポート](https://secon.dev/entry/2024/04/02/080000-japanese-reranker-tech-report/) |
| |
| |
| ## 使い方 |
| |
| ### SentenceTransformers |
| |
| ```python |
| from sentence_transformers import CrossEncoder |
| import torch |
|
|
| MODEL_NAME = "hotchpotch/japanese-reranker-cross-encoder-large-v1" |
| device = "cuda" if torch.cuda.is_available() else "cpu" |
| model = CrossEncoder(MODEL_NAME, max_length=512, device=device) |
| if device == "cuda": |
| model.model.half() |
| query = "感動的な映画について" |
| passages = [ |
| "深いテーマを持ちながらも、観る人の心を揺さぶる名作。登場人物の心情描写が秀逸で、ラストは涙なしでは見られない。", |
| "重要なメッセージ性は評価できるが、暗い話が続くので気分が落ち込んでしまった。もう少し明るい要素があればよかった。", |
| "どうにもリアリティに欠ける展開が気になった。もっと深みのある人間ドラマが見たかった。", |
| "アクションシーンが楽しすぎる。見ていて飽きない。ストーリーはシンプルだが、それが逆に良い。", |
| ] |
| scores = model.predict([(query, passage) for passage in passages]) |
| ``` |
| |
| ## HuggingFace transformers |
|
|
| ```python |
| from transformers import AutoTokenizer, AutoModelForSequenceClassification |
| from torch.nn import Sigmoid |
| |
| MODEL_NAME = "hotchpotch/japanese-reranker-cross-encoder-large-v1" |
| device = "cuda" if torch.cuda.is_available() else "cpu" |
| |
| tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME) |
| model = AutoModelForSequenceClassification.from_pretrained(MODEL_NAME) |
| model.to(device) |
| model.eval() |
| |
| if device == "cuda": |
| model.half() |
| |
| query = "感動的な映画について" |
| passages = [ |
| "深いテーマを持ちながらも、観る人の心を揺さぶる名作。登場人物の心情描写が秀逸で、ラストは涙なしでは見られない。", |
| "重要なメッセージ性は評価できるが、暗い話が続くので気分が落ち込んでしまった。もう少し明るい要素があればよかった。", |
| "どうにもリアリティに欠ける展開が気になった。もっと深みのある人間ドラマが見たかった。", |
| "アクションシーンが楽しすぎる。見ていて飽きない。ストーリーはシンプルだが、それが逆に良い。", |
| ] |
| inputs = tokenizer( |
| [(query, passage) for passage in passages], |
| padding=True, |
| truncation=True, |
| max_length=512, |
| return_tensors="pt", |
| ) |
| inputs = {k: v.to(device) for k, v in inputs.items()} |
| logits = model(**inputs).logits |
| activation = Sigmoid() |
| scores = activation(logits).squeeze().tolist() |
| ``` |
|
|
| ## 評価結果 |
|
|
| | Model Name | [JQaRA](https://huggingface.co/datasets/hotchpotch/JQaRA) | [JaCWIR](https://huggingface.co/datasets/hotchpotch/JaCWIR) | [MIRACL](https://huggingface.co/datasets/miracl/miracl) | [JSQuAD](https://github.com/yahoojapan/JGLUE) | |
| | ------------------------------------------------------------------------------------------------------------------------ | --------------------------------------------------------- | ----------------------------------------------------------- | ------------------------------------------------------- | --------------------------------------------- | |
| | [japanese-reranker-cross-encoder-xsmall-v1](https://huggingface.co/hotchpotch/japanese-reranker-cross-encoder-xsmall-v1) | 0.6136 | 0.9376 | 0.7411 | 0.9602 | |
| | [japanese-reranker-cross-encoder-small-v1](https://huggingface.co/hotchpotch/japanese-reranker-cross-encoder-small-v1) | 0.6247 | 0.939 | 0.7776 | 0.9604 | |
| | [japanese-reranker-cross-encoder-base-v1](https://huggingface.co/hotchpotch/japanese-reranker-cross-encoder-base-v1) | 0.6711 | 0.9337 | 0.818 | 0.9708 | |
| | [japanese-reranker-cross-encoder-large-v1](https://huggingface.co/hotchpotch/japanese-reranker-cross-encoder-large-v1) | 0.7099 | 0.9364 | 0.8406 | 0.9773 | |
| | [japanese-bge-reranker-v2-m3-v1](https://huggingface.co/hotchpotch/japanese-bge-reranker-v2-m3-v1) | 0.6918 | 0.9372 | 0.8423 | 0.9624 | |
| | [bge-reranker-v2-m3](https://huggingface.co/BAAI/bge-reranker-v2-m3) | 0.673 | 0.9343 | 0.8374 | 0.9599 | |
| | [bge-reranker-large](https://huggingface.co/BAAI/bge-reranker-large) | 0.4718 | 0.7332 | 0.7666 | 0.7081 | |
| | [bge-reranker-base](https://huggingface.co/BAAI/bge-reranker-base) | 0.2445 | 0.4905 | 0.6792 | 0.5757 | |
| | [cross-encoder-mmarco-mMiniLMv2-L12-H384-v1](https://huggingface.co/corrius/cross-encoder-mmarco-mMiniLMv2-L12-H384-v1) | 0.5588 | 0.9211 | 0.7158 | 0.932 | |
| | [shioriha-large-reranker](https://huggingface.co/cl-nagoya/shioriha-large-reranker) | 0.5775 | 0.8458 | 0.8084 | 0.9262 | |
| | [bge-m3+all](https://huggingface.co/BAAI/bge-m3) | 0.576 | 0.904 | 0.7926 | 0.9226 | |
| | [bge-m3+dense](https://huggingface.co/BAAI/bge-m3) | 0.539 | 0.8642 | 0.7753 | 0.8815 | |
| | [bge-m3+colbert](https://huggingface.co/BAAI/bge-m3) | 0.5656 | 0.9064 | 0.7902 | 0.9297 | |
| | [bge-m3+sparse](https://huggingface.co/BAAI/bge-m3) | 0.5088 | 0.8944 | 0.6941 | 0.9184 | |
| | [JaColBERTv2](https://huggingface.co/bclavie/JaColBERTv2) | 0.5847 | 0.9185 | 0.6861 | 0.9247 | |
| | [multilingual-e5-large](https://huggingface.co/intfloat/multilingual-e5-large) | 0.554 | 0.8759 | 0.7722 | 0.8892 | |
| | [multilingual-e5-small](https://huggingface.co/intfloat/multilingual-e5-small) | 0.4917 | 0.869 | 0.7025 | 0.8565 | |
| | bm25 | 0.458 | 0.8408 | 0.4387 | 0.9002 | |
|
|
|
|
| ## ライセンス |
|
|
| MIT License |
