dragonkue / multilingual-e5-small-ko

huggingface.co
Total runs: 4.1K
24-hour runs: 259
7-day runs: 907
30-day runs: 1.6K
Model's Last Updated: October 16 2025
sentence-similarity

Introduction of multilingual-e5-small-ko

Model Details of multilingual-e5-small-ko

SentenceTransformer based on intfloat/multilingual-e5-small

This is a sentence-transformers model finetuned from intfloat/multilingual-e5-small on datasets that include Korean query-passage pairs for improved performance on Korean retrieval tasks. It maps sentences & paragraphs to a 384-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.

This model is a lightweight Korean retriever, designed for ease of use and strong performance in practical retrieval tasks. It is ideal for running demos or lightweight applications, offering a good balance between speed and accuracy.

For even higher retrieval performance, we recommend combining it with a reranker. Suggested reranker models:

  • dragonkue/bge-reranker-v2-m3-ko

  • BAAI/bge-reranker-v2-m3

Model Details
Model Description
  • Model Type: Sentence Transformer
  • Base model: intfloat/multilingual-e5-small
  • Maximum Sequence Length: 512 tokens
  • Output Dimensionality: 384 dimensions
  • Similarity Function: Cosine Similarity
  • Training Datasets:
Model Sources
Full Model Architecture
SentenceTransformer(
  (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 384, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
  (2): Normalize()
)
Usage
Direct Usage (Sentence Transformers)

First install the Sentence Transformers library:

pip install -U sentence-transformers

Then you can load this model and run inference.

from sentence_transformers import SentenceTransformer

# Download from the ๐Ÿค— Hub
model = SentenceTransformer("dragonkue/multilingual-e5-small-ko")
# Run inference
sentences = [
    'query: ๋ถํ•œ๊ฐ€์กฑ๋ฒ• ๋ช‡ ์ฐจ ๊ฐœ์ •์—์„œ ์ดํ˜ผํŒ๊ฒฐ ํ™•์ • ํ›„ 3๊ฐœ์›” ๋‚ด์— ๋“ฑ๋ก์‹œ์—๋งŒ ์œ ํšจํ•˜๋‹ค๋Š” ์กฐํ•ญ์„ ํ™•์‹คํžˆ ํ–ˆ์„๊นŒ?',
    'passage: 1990๋…„์— ์ œ์ •๋œ ๋ถํ•œ ๊ฐ€์กฑ๋ฒ•์€ ์ง€๊ธˆ๊นŒ์ง€ 4์ฐจ๋ก€ ๊ฐœ์ •๋˜์–ด ํ˜„์žฌ์— ์ด๋ฅด๊ณ  ์žˆ๋‹ค. 1993๋…„์— ์ด๋ฃจ์–ด์ง„ ์ œ1์ฐจ ๊ฐœ์ •์€ ์ฃผ๋กœ ๊ทœ์ •์˜ ์ •ํ™•์„ฑ์„ ๊ธฐํ•˜๊ธฐ ์œ„ํ•˜์—ฌ ๋ช‡๋ช‡ ์กฐ๋ฌธ์„ ์ˆ˜์ •ํ•œ ๊ฒƒ์ด๋ฉฐ, ์‹ค์ฒด์ ์ธ ๋‚ด์šฉ์„ ๋ณด์™„ํ•œ ๊ฒƒ์€ ์ƒ์†์˜ ์Šน์ธ๊ณผ ํฌ๊ธฐ๊ธฐ๊ฐ„์„ ์„ค์ •ํ•œ ์ œ52์กฐ ์ •๋„๋ผ๊ณ  ํ•  ์ˆ˜ ์žˆ๋‹ค. 2004๋…„์— ์ด๋ฃจ์–ด์ง„ ์ œ2์ฐจ์— ๊ฐœ์ •์—์„œ๋Š” ์ œ20์กฐ์ œ3ํ•ญ์„ ์‹ ์„คํ•˜์—ฌ ์žฌํŒ์ƒ ํ™•์ •๋œ ์ดํ˜ผํŒ๊ฒฐ์„ 3๊ฐœ์›” ๋‚ด์— ๋“ฑ๋กํ•ด์•ผ ์ดํ˜ผ์˜ ํšจ๋ ฅ์ด ๋ฐœ์ƒํ•œ๋‹ค๋Š” ๊ฒƒ์„ ๋ช…ํ™•ํ•˜๊ฒŒ ํ•˜์˜€๋‹ค. 2007๋…„์— ์ด๋ฃจ์–ด์ง„ ์ œ3์ฐจ ๊ฐœ์ •์—์„œ๋Š” ๋ถ€๋ชจ์™€ ์ž๋…€ ๊ด€๊ณ„ ๋˜ํ•œ ์‹ ๋ถ„๋“ฑ๋ก๊ธฐ๊ด€์— ๋“ฑ๋กํ•œ ๋•Œ๋ถ€ํ„ฐ ๋ฒ•์  ํšจ๋ ฅ์ด ๋ฐœ์ƒํ•œ๋‹ค๋Š” ๊ฒƒ์„ ์‹ ์„ค(์ œ25์กฐ์ œ2ํ•ญ)ํ•˜์˜€๋‹ค. ๋˜ํ•œ ๋ฏธ์„ฑ๋…„์ž, ๋…ธ๋™๋Šฅ๋ ฅ ์—†๋Š” ์ž์˜ ๋ถ€์–‘๊ณผ ๊ด€๋ จ(์ œ37์กฐ์ œ2ํ•ญ)ํ•˜์—ฌ ๊ธฐ์กด์—๋Š” โ€œ๋ถ€์–‘๋Šฅ๋ ฅ์ด ์žˆ๋Š” ๊ฐ€์ •์„ฑ์›์ด ์—†์„ ๊ฒฝ์šฐ์—๋Š” ๋”ฐ๋กœ ์‚ฌ๋Š” ๋ถ€๋ชจ๋‚˜ ์ž๋…€, ์กฐ๋ถ€๋ชจ๋‚˜ ์†์ž๋…€, ํ˜•์ œ์ž๋งค๊ฐ€ ๋ถ€์–‘ํ•œ๋‹คโ€๊ณ  ๊ทœ์ •ํ•˜๊ณ  ์žˆ์—ˆ๋˜ ๊ฒƒ์„ โ€œ๋ถ€์–‘๋Šฅ๋ ฅ์ด ์žˆ๋Š” ๊ฐ€์ •์„ฑ์›์ด ์—†์„ ๊ฒฝ์šฐ์—๋Š” ๋”ฐ๋กœ ์‚ฌ๋Š” ๋ถ€๋ชจ๋‚˜ ์ž๋…€๊ฐ€ ๋ถ€์–‘ํ•˜๋ฉฐ ๊ทธ๋“ค์ด ์—†์„ ๊ฒฝ์šฐ์—๋Š” ์กฐ๋ถ€๋ชจ๋‚˜ ์†์ž๋…€, ํ˜•์ œ์ž๋งค๊ฐ€ ๋ถ€์–‘ํ•œ๋‹คโ€๋กœ ๊ฐœ์ •ํ•˜์˜€๋‹ค.',
    'passage: ํ™˜๊ฒฝ๋งˆํฌ ์ œ๋„, ์ธ์ฆ๊ธฐ์ค€ ๋ณ€๊ฒฝ์œผ๋กœ ๊ธฐ์—…๋ถ€๋‹ด ์ค„์ธ๋‹ค\nํ™˜๊ฒฝ๋งˆํฌ ์ œ๋„ ์†Œ๊ฐœ\nโ–ก ๊ฐœ์š”\nโ—‹ ๋™์ผ ์šฉ๋„์˜ ๋‹ค๋ฅธ ์ œํ’ˆ์— ๋น„ํ•ด โ€˜์ œํ’ˆ์˜ ํ™˜๊ฒฝ์„ฑ*โ€™์„ ๊ฐœ์„ ํ•œ ์ œํ’ˆ์— ๋กœ๊ณ ์™€ ์„ค๋ช…์„ ํ‘œ์‹œํ•  ์ˆ˜ ์žˆ๋„๋กํ•˜๋Š” ์ธ์ฆ ์ œ๋„\nโ€ป ์ œํ’ˆ์˜ ํ™˜๊ฒฝ์„ฑ : ์žฌ๋ฃŒ์™€ ์ œํ’ˆ์„ ์ œ์กฐโ€ค์†Œ๋น„ ํ๊ธฐํ•˜๋Š” ์ „๊ณผ์ •์—์„œ ์˜ค์—ผ๋ฌผ์งˆ์ด๋‚˜ ์˜จ์‹ค๊ฐ€์Šค ๋“ฑ์„ ๋ฐฐ์ถœํ•˜๋Š” ์ •๋„ ๋ฐ ์ž์›๊ณผ ์—๋„ˆ์ง€๋ฅผ ์†Œ๋น„ํ•˜๋Š” ์ •๋„ ๋“ฑ ํ™˜๊ฒฝ์— ๋ฏธ์น˜๋Š” ์˜ํ–ฅ๋ ฅ์˜ ์ •๋„(ใ€Œํ™˜๊ฒฝ๊ธฐ์ˆ  ๋ฐ ํ™˜๊ฒฝ์‚ฐ์—… ์ง€์›๋ฒ•ใ€์ œ2์กฐ์ œ5ํ˜ธ)\nโ–ก ๋ฒ•์ ๊ทผ๊ฑฐ\nโ—‹ ใ€Œํ™˜๊ฒฝ๊ธฐ์ˆ  ๋ฐ ํ™˜๊ฒฝ์‚ฐ์—… ์ง€์›๋ฒ•ใ€์ œ17์กฐ(ํ™˜๊ฒฝํ‘œ์ง€์˜ ์ธ์ฆ)\nโ–ก ๊ด€๋ จ ๊ตญ์ œํ‘œ์ค€\nโ—‹ ISO 14024(์ œ1์œ ํ˜• ํ™˜๊ฒฝ๋ผ๋ฒจ๋ง)\nโ–ก ์ ์šฉ๋Œ€์ƒ\nโ—‹ ์‚ฌ๋ฌด๊ธฐ๊ธฐ, ๊ฐ€์ „์ œํ’ˆ, ์ƒํ™œ์šฉํ’ˆ, ๊ฑด์ถ•์ž์žฌ ๋“ฑ 156๊ฐœ ๋Œ€์ƒ์ œํ’ˆ๊ตฐ\nโ–ก ์ธ์ฆํ˜„ํ™ฉ\nโ—‹ 2,737๊ฐœ ๊ธฐ์—…์˜ 16,647๊ฐœ ์ œํ’ˆ(2015.12์›”๋ง ๊ธฐ์ค€)',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 384]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]
Direct Usage (Transformers)
import torch.nn.functional as F

from torch import Tensor
from transformers import AutoTokenizer, AutoModel


def average_pool(last_hidden_states: Tensor,
                 attention_mask: Tensor) -> Tensor:
    last_hidden = last_hidden_states.masked_fill(~attention_mask[..., None].bool(), 0.0)
    return last_hidden.sum(dim=1) / attention_mask.sum(dim=1)[..., None]


# Each input text should start with "query: " or "passage: ", even for non-English texts.
# For tasks other than retrieval, you can simply use the "query: " prefix.
input_texts = ["query: ๋ถํ•œ๊ฐ€์กฑ๋ฒ• ๋ช‡ ์ฐจ ๊ฐœ์ •์—์„œ ์ดํ˜ผํŒ๊ฒฐ ํ™•์ • ํ›„ 3๊ฐœ์›” ๋‚ด์— ๋“ฑ๋ก์‹œ์—๋งŒ ์œ ํšจํ•˜๋‹ค๋Š” ์กฐํ•ญ์„ ํ™•์‹คํžˆ ํ–ˆ์„๊นŒ?",
               "passage: 1990๋…„์— ์ œ์ •๋œ ๋ถํ•œ ๊ฐ€์กฑ๋ฒ•์€ ์ง€๊ธˆ๊นŒ์ง€ 4์ฐจ๋ก€ ๊ฐœ์ •๋˜์–ด ํ˜„์žฌ์— ์ด๋ฅด๊ณ  ์žˆ๋‹ค. 1993๋…„์— ์ด๋ฃจ์–ด์ง„ ์ œ1์ฐจ ๊ฐœ์ •์€ ์ฃผ๋กœ ๊ทœ์ •์˜ ์ •ํ™•์„ฑ์„ ๊ธฐํ•˜๊ธฐ ์œ„ํ•˜์—ฌ ๋ช‡๋ช‡ ์กฐ๋ฌธ์„ ์ˆ˜์ •ํ•œ ๊ฒƒ์ด๋ฉฐ, ์‹ค์ฒด์ ์ธ ๋‚ด์šฉ์„ ๋ณด์™„ํ•œ ๊ฒƒ์€ ์ƒ์†์˜ ์Šน์ธ๊ณผ ํฌ๊ธฐ๊ธฐ๊ฐ„์„ ์„ค์ •ํ•œ ์ œ52์กฐ ์ •๋„๋ผ๊ณ  ํ•  ์ˆ˜ ์žˆ๋‹ค. 2004๋…„์— ์ด๋ฃจ์–ด์ง„ ์ œ2์ฐจ์— ๊ฐœ์ •์—์„œ๋Š” ์ œ20์กฐ์ œ3ํ•ญ์„ ์‹ ์„คํ•˜์—ฌ ์žฌํŒ์ƒ ํ™•์ •๋œ ์ดํ˜ผํŒ๊ฒฐ์„ 3๊ฐœ์›” ๋‚ด์— ๋“ฑ๋กํ•ด์•ผ ์ดํ˜ผ์˜ ํšจ๋ ฅ์ด ๋ฐœ์ƒํ•œ๋‹ค๋Š” ๊ฒƒ์„ ๋ช…ํ™•ํ•˜๊ฒŒ ํ•˜์˜€๋‹ค. 2007๋…„์— ์ด๋ฃจ์–ด์ง„ ์ œ3์ฐจ ๊ฐœ์ •์—์„œ๋Š” ๋ถ€๋ชจ์™€ ์ž๋…€ ๊ด€๊ณ„ ๋˜ํ•œ ์‹ ๋ถ„๋“ฑ๋ก๊ธฐ๊ด€์— ๋“ฑ๋กํ•œ ๋•Œ๋ถ€ํ„ฐ ๋ฒ•์  ํšจ๋ ฅ์ด ๋ฐœ์ƒํ•œ๋‹ค๋Š” ๊ฒƒ์„ ์‹ ์„ค(์ œ25์กฐ์ œ2ํ•ญ)ํ•˜์˜€๋‹ค. ๋˜ํ•œ ๋ฏธ์„ฑ๋…„์ž, ๋…ธ๋™๋Šฅ๋ ฅ ์—†๋Š” ์ž์˜ ๋ถ€์–‘๊ณผ ๊ด€๋ จ(์ œ37์กฐ์ œ2ํ•ญ)ํ•˜์—ฌ ๊ธฐ์กด์—๋Š” โ€œ๋ถ€์–‘๋Šฅ๋ ฅ์ด ์žˆ๋Š” ๊ฐ€์ •์„ฑ์›์ด ์—†์„ ๊ฒฝ์šฐ์—๋Š” ๋”ฐ๋กœ ์‚ฌ๋Š” ๋ถ€๋ชจ๋‚˜ ์ž๋…€, ์กฐ๋ถ€๋ชจ๋‚˜ ์†์ž๋…€, ํ˜•์ œ์ž๋งค๊ฐ€ ๋ถ€์–‘ํ•œ๋‹คโ€๊ณ  ๊ทœ์ •ํ•˜๊ณ  ์žˆ์—ˆ๋˜ ๊ฒƒ์„ โ€œ๋ถ€์–‘๋Šฅ๋ ฅ์ด ์žˆ๋Š” ๊ฐ€์ •์„ฑ์›์ด ์—†์„ ๊ฒฝ์šฐ์—๋Š” ๋”ฐ๋กœ ์‚ฌ๋Š” ๋ถ€๋ชจ๋‚˜ ์ž๋…€๊ฐ€ ๋ถ€์–‘ํ•˜๋ฉฐ ๊ทธ๋“ค์ด ์—†์„ ๊ฒฝ์šฐ์—๋Š” ์กฐ๋ถ€๋ชจ๋‚˜ ์†์ž๋…€, ํ˜•์ œ์ž๋งค๊ฐ€ ๋ถ€์–‘ํ•œ๋‹คโ€๋กœ ๊ฐœ์ •ํ•˜์˜€๋‹ค.",
               "passage: ํ™˜๊ฒฝ๋งˆํฌ ์ œ๋„, ์ธ์ฆ๊ธฐ์ค€ ๋ณ€๊ฒฝ์œผ๋กœ ๊ธฐ์—…๋ถ€๋‹ด ์ค„์ธ๋‹ค\nํ™˜๊ฒฝ๋งˆํฌ ์ œ๋„ ์†Œ๊ฐœ\nโ–ก ๊ฐœ์š”\nโ—‹ ๋™์ผ ์šฉ๋„์˜ ๋‹ค๋ฅธ ์ œํ’ˆ์— ๋น„ํ•ด โ€˜์ œํ’ˆ์˜ ํ™˜๊ฒฝ์„ฑ*โ€™์„ ๊ฐœ์„ ํ•œ ์ œํ’ˆ์— ๋กœ๊ณ ์™€ ์„ค๋ช…์„ ํ‘œ์‹œํ•  ์ˆ˜ ์žˆ๋„๋กํ•˜๋Š” ์ธ์ฆ ์ œ๋„\nโ€ป ์ œํ’ˆ์˜ ํ™˜๊ฒฝ์„ฑ : ์žฌ๋ฃŒ์™€ ์ œํ’ˆ์„ ์ œ์กฐโ€ค์†Œ๋น„ ํ๊ธฐํ•˜๋Š” ์ „๊ณผ์ •์—์„œ ์˜ค์—ผ๋ฌผ์งˆ์ด๋‚˜ ์˜จ์‹ค๊ฐ€์Šค ๋“ฑ์„ ๋ฐฐ์ถœํ•˜๋Š” ์ •๋„ ๋ฐ ์ž์›๊ณผ ์—๋„ˆ์ง€๋ฅผ ์†Œ๋น„ํ•˜๋Š” ์ •๋„ ๋“ฑ ํ™˜๊ฒฝ์— ๋ฏธ์น˜๋Š” ์˜ํ–ฅ๋ ฅ์˜ ์ •๋„(ใ€Œํ™˜๊ฒฝ๊ธฐ์ˆ  ๋ฐ ํ™˜๊ฒฝ์‚ฐ์—… ์ง€์›๋ฒ•ใ€์ œ2์กฐ์ œ5ํ˜ธ)\nโ–ก ๋ฒ•์ ๊ทผ๊ฑฐ\nโ—‹ ใ€Œํ™˜๊ฒฝ๊ธฐ์ˆ  ๋ฐ ํ™˜๊ฒฝ์‚ฐ์—… ์ง€์›๋ฒ•ใ€์ œ17์กฐ(ํ™˜๊ฒฝํ‘œ์ง€์˜ ์ธ์ฆ)\nโ–ก ๊ด€๋ จ ๊ตญ์ œํ‘œ์ค€\nโ—‹ ISO 14024(์ œ1์œ ํ˜• ํ™˜๊ฒฝ๋ผ๋ฒจ๋ง)\nโ–ก ์ ์šฉ๋Œ€์ƒ\nโ—‹ ์‚ฌ๋ฌด๊ธฐ๊ธฐ, ๊ฐ€์ „์ œํ’ˆ, ์ƒํ™œ์šฉํ’ˆ, ๊ฑด์ถ•์ž์žฌ ๋“ฑ 156๊ฐœ ๋Œ€์ƒ์ œํ’ˆ๊ตฐ\nโ–ก ์ธ์ฆํ˜„ํ™ฉ\nโ—‹ 2,737๊ฐœ ๊ธฐ์—…์˜ 16,647๊ฐœ ์ œํ’ˆ(2015.12์›”๋ง ๊ธฐ์ค€)"]

tokenizer = AutoTokenizer.from_pretrained('dragonkue/multilingual-e5-small-ko')
model = AutoModel.from_pretrained('dragonkue/multilingual-e5-small-ko')

# Tokenize the input texts
batch_dict = tokenizer(input_texts, max_length=512, padding=True, truncation=True, return_tensors='pt')

outputs = model(**batch_dict)
embeddings = average_pool(outputs.last_hidden_state, batch_dict['attention_mask'])

# normalize embeddings
embeddings = F.normalize(embeddings, p=2, dim=1)
scores = (embeddings[:1] @ embeddings[1:].T)
print(scores.tolist())
Evaluation
Korean Retrieval Benchmark
  • Ko-StrategyQA : A Korean ODQA multi-hop retrieval dataset , translated from StrategyQA.
  • AutoRAGRetrieval : A Korean document retrieval dataset constructed by parsing PDFs from five domains: finance, public, medical, legal, and commerce .
  • MIRACLRetrieval : A Korean document retrieval dataset based on Wikipedia.
  • PublicHealthQA : A retrieval dataset focused on medical and public health domains in Korean.
  • BelebeleRetrieval : A Korean document retrieval dataset based on FLORES-200.
  • MrTidyRetrieval : A Wikipedia-based Korean document retrieval dataset .
  • XPQARetrieval : A cross-domain Korean document retrieval dataset .
Metrics
  • Standard metric : NDCG@10
Information Retrieval
Model Size(M) Average XPQARetrieval PublicHealthQA MIRACLRetrieval Ko-StrategyQA BelebeleRetrieval AutoRAGRetrieval MrTidyRetrieval
BAAI/bge-m3 560 0.724169 0.36075 0.80412 0.70146 0.79405 0.93164 0.83008 0.64708
Snowflake/snowflake-arctic-embed-l-v2.0 560 0.724104 0.43018 0.81679 0.66077 0.80455 0.9271 0.83863 0.59071
intfloat/multilingual-e5-large 560 0.721607 0.3571 0.82534 0.66486 0.80348 0.94499 0.81337 0.64211
intfloat/multilingual-e5-base 278 0.689429 0.3607 0.77203 0.6227 0.76355 0.92868 0.79752 0.58082
dragonkue/multilingual-e5-small-ko 118 0.688819 0.34871 0.79729 0.61113 0.76173 0.9297 0.86184 0.51133
intfloat/multilingual-e5-small 118 0.670906 0.33003 0.73668 0.61238 0.75157 0.90531 0.80068 0.55969
ibm-granite/granite-embedding-278m-multilingual 278 0.616466 0.23058 0.77668 0.59216 0.71762 0.83231 0.70226 0.46365
ibm-granite/granite-embedding-107m-multilingual 107 0.599759 0.23058 0.73209 0.58413 0.70531 0.82063 0.68243 0.44314
sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2 118 0.409766 0.21345 0.67409 0.25676 0.45903 0.71491 0.42296 0.12716
Performance Comparison by Model Size (Based on Average NDCG@10)
Training Details
Training Datasets

This model was fine-tuned on the same dataset used in dragonkue/snowflake-arctic-embed-l-v2.0-ko, which consists of Korean query-passage pairs. The training objective was to improve retrieval performance specifically for Korean-language tasks.

Training Methods

Following the training approach used in dragonkue/snowflake-arctic-embed-l-v2.0-ko, this model constructs in-batch negatives based on clustered passages. In addition, we introduce GISTEmbedLoss with a configurable margin.

๐Ÿ“ˆ Margin-based Training Results

  • Using the standard MNR (Multiple Negatives Ranking) loss alone resulted in decreased performance.

  • The original GISTEmbedLoss (without margin) yielded modest improvements of around +0.8 NDCG@10.

  • Applying a margin led to performance gains of up to +1.5 NDCG@10.

  • This indicates that simply tuning the margin value can lead to up to 2x improvement, showing strong sensitivity and effectiveness of margin scaling.

This margin-based approach extends the idea proposed in the NV-Retriever paper, which originally filtered false negatives during hard negative sampling. We adapt this to in-batch negatives, treating false negatives as dynamic samples guided by margin-based filtering.

The sentence-transformers library now supports GISTEmbedLoss with margin configuration, making it easy to integrate into any training pipeline.

You can install the latest version with:

pip install -U sentence-transformers
Training Hyperparameters
Non-Default Hyperparameters
  • eval_strategy : steps
  • per_device_train_batch_size : 20000
  • per_device_eval_batch_size : 4096
  • learning_rate : 0.00025
  • num_train_epochs : 3
  • warmup_ratio : 0.05
  • fp16 : True
  • dataloader_drop_last : True
  • batch_sampler : no_duplicates
All Hyperparameters
Click to expand
  • overwrite_output_dir : False
  • do_predict : False
  • eval_strategy : steps
  • prediction_loss_only : True
  • per_device_train_batch_size : 20000
  • per_device_eval_batch_size : 4096
  • per_gpu_train_batch_size : None
  • per_gpu_eval_batch_size : None
  • gradient_accumulation_steps : 1
  • eval_accumulation_steps : None
  • torch_empty_cache_steps : None
  • learning_rate : 0.00025
  • weight_decay : 0.0
  • adam_beta1 : 0.9
  • adam_beta2 : 0.999
  • adam_epsilon : 1e-08
  • max_grad_norm : 1.0
  • num_train_epochs : 2
  • max_steps : -1
  • lr_scheduler_type : linear
  • lr_scheduler_kwargs : {}
  • warmup_ratio : 0.05
  • warmup_steps : 0
  • log_level : passive
  • log_level_replica : warning
  • log_on_each_node : True
  • logging_nan_inf_filter : True
  • save_safetensors : True
  • save_on_each_node : False
  • save_only_model : False
  • restore_callback_states_from_checkpoint : False
  • no_cuda : False
  • use_cpu : False
  • use_mps_device : False
  • seed : 42
  • data_seed : None
  • jit_mode_eval : False
  • use_ipex : False
  • bf16 : False
  • fp16 : True
  • fp16_opt_level : O1
  • half_precision_backend : auto
  • bf16_full_eval : False
  • fp16_full_eval : False
  • tf32 : None
  • local_rank : 0
  • ddp_backend : None
  • tpu_num_cores : None
  • tpu_metrics_debug : False
  • debug : []
  • dataloader_drop_last : True
  • dataloader_num_workers : 0
  • dataloader_prefetch_factor : None
  • past_index : -1
  • disable_tqdm : False
  • remove_unused_columns : True
  • label_names : None
  • load_best_model_at_end : False
  • ignore_data_skip : False
  • fsdp : []
  • fsdp_min_num_params : 0
  • fsdp_config : {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
  • tp_size : 0
  • fsdp_transformer_layer_cls_to_wrap : None
  • accelerator_config : {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
  • deepspeed : None
  • label_smoothing_factor : 0.0
  • optim : adamw_torch
  • optim_args : None
  • adafactor : False
  • group_by_length : False
  • length_column_name : length
  • ddp_find_unused_parameters : None
  • ddp_bucket_cap_mb : None
  • ddp_broadcast_buffers : False
  • dataloader_pin_memory : True
  • dataloader_persistent_workers : False
  • skip_memory_metrics : True
  • use_legacy_prediction_loop : False
  • push_to_hub : False
  • resume_from_checkpoint : None
  • hub_model_id : None
  • hub_strategy : every_save
  • hub_private_repo : None
  • hub_always_push : False
  • gradient_checkpointing : False
  • gradient_checkpointing_kwargs : None
  • include_inputs_for_metrics : False
  • include_for_metrics : []
  • eval_do_concat_batches : True
  • fp16_backend : auto
  • push_to_hub_model_id : None
  • push_to_hub_organization : None
  • mp_parameters :
  • auto_find_batch_size : False
  • full_determinism : False
  • torchdynamo : None
  • ray_scope : last
  • ddp_timeout : 1800
  • torch_compile : False
  • torch_compile_backend : None
  • torch_compile_mode : None
  • include_tokens_per_second : False
  • include_num_input_tokens_seen : False
  • neftune_noise_alpha : None
  • optim_target_modules : None
  • batch_eval_metrics : False
  • eval_on_start : False
  • use_liger_kernel : False
  • eval_use_gather_object : False
  • average_tokens_across_devices : False
  • prompts : None
  • batch_sampler : no_duplicates
  • multi_dataset_batch_sampler : proportional
Framework Versions
  • Python: 3.11.10
  • Sentence Transformers: 4.1.0
  • Transformers: 4.51.3
  • PyTorch: 2.7.0+cu126
  • Accelerate: 1.6.0
  • Datasets: 3.5.1
  • Tokenizers: 0.21.1
FAQ

1. Do I need to add the prefix "query: " and "passage: " to input texts?

Yes, this is how the model is trained, otherwise you will see a performance degradation.

Here are some rules of thumb:

Use "query: " and "passage: " correspondingly for asymmetric tasks such as passage retrieval in open QA, ad-hoc information retrieval.

Use "query: " prefix for symmetric tasks such as semantic similarity, bitext mining, paraphrase retrieval.

Use "query: " prefix if you want to use embeddings as features, such as linear probing classification, clustering.

2. Why does the cosine similarity scores distribute around 0.7 to 1.0?

This is a known and expected behavior as we use a low temperature 0.01 for InfoNCE contrastive loss.

For text embedding tasks like text retrieval or semantic similarity, what matters is the relative order of the scores instead of the absolute values, so this should not be an issue.

Citation
BibTeX
Sentence Transformers
@inproceedings{reimers-2019-sentence-bert,
    title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
    author = "Reimers, Nils and Gurevych, Iryna",
    booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
    month = "11",
    year = "2019",
    publisher = "Association for Computational Linguistics",
    url = "https://arxiv.org/abs/1908.10084",
}
Base Model
@article{wang2024multilingual,
  title={Multilingual E5 Text Embeddings: A Technical Report},
  author={Wang, Liang and Yang, Nan and Huang, Xiaolong and Yang, Linjun and Majumder, Rangan and Wei, Furu},
  journal={arXiv preprint arXiv:2402.05672},
  year={2024}
}
NV-Retriever: Improving text embedding models with effective hard-negative mining
@article{moreira2024nvretriever,
  title     = {NV-Retriever: Improving text embedding models with effective hard-negative mining},
  author    = {Moreira, Gabriel de Souza P. and Osmulski, Radek and Xu, Mengyao and Ak, Ronay and Schifferer, Benedikt and Oldridge, Even},
  journal   = {arXiv preprint arXiv:2407.15831},
  year      = {2024},
  url       = {https://arxiv.org/abs/2407.15831},
  doi       = {10.48550/arXiv.2407.15831}
}
Limitations

Long texts will be truncated to at most 512 tokens.

Runs of dragonkue multilingual-e5-small-ko on huggingface.co

4.1K
Total runs
259
24-hour runs
674
3-day runs
907
7-day runs
1.6K
30-day runs

More Information About multilingual-e5-small-ko huggingface.co Model

More multilingual-e5-small-ko license Visit here:

https://choosealicense.com/licenses/apache-2.0

multilingual-e5-small-ko huggingface.co

multilingual-e5-small-ko huggingface.co is an AI model on huggingface.co that provides multilingual-e5-small-ko's model effect (), which can be used instantly with this dragonkue multilingual-e5-small-ko model. huggingface.co supports a free trial of the multilingual-e5-small-ko model, and also provides paid use of the multilingual-e5-small-ko. Support call multilingual-e5-small-ko model through api, including Node.js, Python, http.

multilingual-e5-small-ko huggingface.co Url

https://huggingface.co/dragonkue/multilingual-e5-small-ko

dragonkue multilingual-e5-small-ko online free

multilingual-e5-small-ko huggingface.co is an online trial and call api platform, which integrates multilingual-e5-small-ko's modeling effects, including api services, and provides a free online trial of multilingual-e5-small-ko, you can try multilingual-e5-small-ko online for free by clicking the link below.

dragonkue multilingual-e5-small-ko online free url in huggingface.co:

https://huggingface.co/dragonkue/multilingual-e5-small-ko

multilingual-e5-small-ko install

multilingual-e5-small-ko is an open source model from GitHub that offers a free installation service, and any user can find multilingual-e5-small-ko on GitHub to install. At the same time, huggingface.co provides the effect of multilingual-e5-small-ko install, users can directly use multilingual-e5-small-ko installed effect in huggingface.co for debugging and trial. It also supports api for free installation.

multilingual-e5-small-ko install url in huggingface.co:

https://huggingface.co/dragonkue/multilingual-e5-small-ko

Url of multilingual-e5-small-ko

multilingual-e5-small-ko huggingface.co Url

Provider of multilingual-e5-small-ko huggingface.co

dragonkue
ORGANIZATIONS

Other API from dragonkue

huggingface.co

Total runs: 76.5K
Run Growth: -183.8K
Growth Rate: -240.24%
Updated:October 16 2025