cl-nagoya/ruri-v3-dataset-ft
Viewer • Updated • 632k • 383 • 4
How to use iamtatsuki05/Sentence-Llama-Bi-JP-1B with sentence-transformers:
from sentence_transformers import SentenceTransformer
model = SentenceTransformer("iamtatsuki05/Sentence-Llama-Bi-JP-1B", trust_remote_code=True)
sentences = [
"The weather is lovely today.",
"It's so sunny outside!",
"He drove to the stadium."
]
embeddings = model.encode(sentences)
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]How to use iamtatsuki05/Sentence-Llama-Bi-JP-1B with Transformers:
# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModel
tokenizer = AutoTokenizer.from_pretrained("iamtatsuki05/Sentence-Llama-Bi-JP-1B", trust_remote_code=True)
model = AutoModel.from_pretrained("iamtatsuki05/Sentence-Llama-Bi-JP-1B", trust_remote_code=True, device_map="auto")English / Japanese
Sentence-Llama-Bi-JP-1B fine-tunes iamtatsuki05/Sentence-Llama-Bi-JP-1B-PT with supervised contrastive learning on cl-nagoya/ruri-v3-dataset-ft, yielding 1,792-dimensional Japanese sentence embeddings.
sentence-transformers>=4.1.0
transformers>=4.51.0
accelerate>=1.6.0
sentencepiece>=0.2.0
flash-attn>=2.7.3
import torch
from sentence_transformers import SentenceTransformer
model_name = "iamtatsuki05/Sentence-Llama-Bi-JP-1B"
model_kwargs = {
"torch_dtype": torch.bfloat16,
"attn_implementation": "flash_attention_2",
}
model = SentenceTransformer(model_name, model_kwargs=model_kwargs, trust_remote_code=True)
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)
These encoders start from weakly supervised checkpoints and undergo supervised fine-tuning on cl-nagoya/ruri-v3-dataset-ft.
| ID | Architecture | #Param. | #Param. w/o Emb. |
JMTEB-Avg | JMTEB-Retrieval | JMTEB-STS | JMTEB-Classification | JMTEB-Reranking | JMTEB-Clustering |
|---|---|---|---|---|---|---|---|---|---|
| iamtatsuki05/Sentence-ModernBERT-JP-1B | ModernBERT | 1304M | 1120M | 63.51 | 60.52 | 81.61 | 57.92 | 67.04 | 50.47 |
| iamtatsuki05/Sentence-Llama-Bi-JP-1B (this model) |
Llama | 1224M | 1040M | 61.50 | 56.38 | 79.14 | 55.63 | 64.66 | 51.70 |
| iamtatsuki05/Sentence-Sarashina-Bi-JP-1B | Llama | 1224M | 1040M | 66.36 | 65.03 | 83.16 | 61.90 | 71.37 | 50.34 |
Scores are measured on JMTEB v2. Each task-family score averages all datasets, splits, and subsets in the family, and JMTEB-Avg is the macro average over the five families.
This model is distributed under the MIT License.
@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}
}
Base model
iamtatsuki05/Llama-JP-1B-init