---
language:
- ja
license: mit
library_name: sentence-transformers
pipeline_tag: sentence-similarity
tags:
- mirei
- modernbert
- masked-lm
- transformers
- text-embedding
- embeddings
- feature-extraction
- retrieval
base_model: iamtatsuki05/ModernBERT-JP-0.5B-PT-stage2
datasets:
- cl-nagoya/ruri-dataset-v2-pt
---

# Sentence-ModernBERT-JP-0.5B-PT

English / [Japanese](README_JA.md)


## Overview
Sentence-ModernBERT-JP-0.5B-PT wraps [iamtatsuki05/ModernBERT-JP-0.5B-PT-stage2](https://huggingface.co/iamtatsuki05/ModernBERT-JP-0.5B-PT-stage2) with weakly supervised contrastive training on [cl-nagoya/ruri-dataset-v2-pt](https://huggingface.co/datasets/cl-nagoya/ruri-dataset-v2-pt), producing 1,280-dimensional Japanese sentence embeddings.


- **[Hugging Face Collection](https://huggingface.co/collections/iamtatsuki05/mirei)**
- **[GitHub](https://github.com/iamtatsuki05/MIREI)**

![Consept](assets/concept.jpg)

## Usage
### Requirements
```
sentence-transformers>=4.1.0
transformers>=4.51.0
accelerate>=1.6.0
sentencepiece>=0.2.0
flash-attn>=2.7.3
```

### Sample Code
```python
import torch
from sentence_transformers import SentenceTransformer

model_name = "iamtatsuki05/Sentence-ModernBERT-JP-0.5B-PT"
model_kwargs = {
  "torch_dtype": torch.bfloat16,
  "attn_implementation": "flash_attention_2",
}
model = SentenceTransformer(model_name, model_kwargs=model_kwargs)

queries = ["ハチワレはどのようなキャラクターですか？"]
docs = [
    "ハチワレは、『ちいかわ』に登場する猫風のキャラクターで、明るく社交的、前向きな性格が特徴。ちいかわたちと共に日常を楽しみつつ、討伐などの冒険にも積極的に挑む存在です。",
    "うさぎは、天真爛漫でマイペースな性格が特徴のキャラクターで、突飛な行動力と鋭い直感でちいかわたちを引っ張る存在。自由気ままながらも仲間思いな一面を併せ持ちます。",
]
q_emb = model.encode(queries, normalize_embeddings=True)
d_emb = model.encode(docs, normalize_embeddings=True)
scores = model.similarity(q_emb, d_emb)
print(scores)
```


## Model Details
- **Base model:** [iamtatsuki05/ModernBERT-JP-0.5B-PT-stage2](https://huggingface.co/iamtatsuki05/ModernBERT-JP-0.5B-PT-stage2)
- **Architecture:** ModernBERT
- **Maximum sequence length:** 8,192 tokens
- **Embedding dimension:** 1280 (mean pooling)
- **Tokenizer:** SentencePiece / vocabulary size 102,400
- **Positional encoding:** RoPE
- **Supported languages:** Japanese
- **Similarity metric:** cosine


## Model Series

The following encoders share the same weakly supervised recipe on [cl-nagoya/ruri-dataset-v2-pt](https://huggingface.co/datasets/cl-nagoya/ruri-dataset-v2-pt), using roughly two million pairs per subset.

| ID                                                                                                                | Architecture | #Param. | #Param.<br>w/o Emb. |
|:-:|:-:|:-:|:-:|
| [iamtatsuki05/Sentence-ModernBERT-JP-0.5B-PT](https://huggingface.co/iamtatsuki05/Sentence-ModernBERT-JP-0.5B-PT)<br>(this model) | ModernBERT   | 679M    | 548M                |
| [iamtatsuki05/Sentence-Llama-Bi-JP-0.5B-PT](https://huggingface.co/iamtatsuki05/Sentence-Llama-Bi-JP-0.5B-PT)     | Llama        | 661M    | 530M                |
| [iamtatsuki05/Sentence-Sarashina-Bi-0.5B-PT](https://huggingface.co/iamtatsuki05/Sentence-Sarashina-Bi-0.5B-PT)   | Llama        | 661M    | 530M                |


## Licence
This model is distributed under the [MIT License](https://opensource.org/license/mit/).


## How to Cite

```tex
@article{MIREI
  title={同一条件下における Encoder/Decoder アーキテクチャによる文埋め込みの性能分析},
  author={岡田 龍樹 and 杉本 徹},
  journal={言語処理学会第 32 回年次大会 (NLP2026)},
  year={2026}
}
```