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

Documentation: Sentence Transformers Documentation
Repository: Sentence Transformers on GitHub
Hugging Face: Sentence Transformers on Hugging Face

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

🪶 Lightweight Version Available

We also introduce a lightweight variant of this model: exp-models/dragonkue-KoEn-E5-Tiny, which removes all tokens except Korean and English to reduce model size while maintaining performance.

The repository also includes a GGUF-quantized version, making it suitable for efficient local or on-device embedding model serving.

🔧 For practical deployment, we highly recommend using this lightweight retriever in combination with a reranker model — it forms a powerful and resource-efficient retrieval setup.

Direct Usage (Sentence Transformers)

First install the Sentence Transformers library:

pip install -U sentence-transformers

Then you can load this model and run inference.

python

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("dragonkue/multilingual-e5-small-ko")
# Run inference
sentences = [
    'query: 북한가족법 몇 차 개정에서 이혼판결 확정 후 3개월 내에 등록시에만 유효하다는 조항을 확실히 했을까?',
    'passage: 1990년에 제정된 북한 가족법은 지금까지 4차례 개정되어 현재에 이르고 있다. 1993년에 이루어진 제1차 개정은 주로 규정의 정확성을 기하기 위하여 몇몇 조문을 수정한 것이며, 실체적인 내용을 보완한 것은 상속의 승인과 포기기간을 설정한 제52조 정도라고 할 수 있다. 2004년에 이루어진 제2차에 개정에서는 제20조제3항을 신설하여 재판상 확정된 이혼판결을 3개월 내에 등록해야 이혼의 효력이 발생한다는 것을 명확하게 하였다. 2007년에 이루어진 제3차 개정에서는 부모와 자녀 관계 또한 신분등록기관에 등록한 때부터 법적 효력이 발생한다는 것을 신설(제25조제2항)하였다. 또한 미성년자, 노동능력 없는 자의 부양과 관련(제37조제2항)하여 기존에는 “부양능력이 있는 가정성원이 없을 경우에는 따로 사는 부모나 자녀, 조부모나 손자녀, 형제자매가 부양한다”고 규정하고 있었던 것을 “부양능력이 있는 가정성원이 없을 경우에는 따로 사는 부모나 자녀가 부양하며 그들이 없을 경우에는 조부모나 손자녀, 형제자매가 부양한다”로 개정하였다.',
    'passage: 환경마크 제도, 인증기준 변경으로 기업부담 줄인다\n환경마크 제도 소개\n□ 개요\n○ 동일 용도의 다른 제품에 비해 ‘제품의 환경성*’을 개선한 제품에 로고와 설명을 표시할 수 있도록하는 인증 제도\n※ 제품의 환경성 : 재료와 제품을 제조․소비 폐기하는 전과정에서 오염물질이나 온실가스 등을 배출하는 정도 및 자원과 에너지를 소비하는 정도 등 환경에 미치는 영향력의 정도(「환경기술 및 환경산업 지원법」제2조제5호)\n□ 법적근거\n○ 「환경기술 및 환경산업 지원법」제17조(환경표지의 인증)\n□ 관련 국제표준\n○ ISO 14024(제1유형 환경라벨링)\n□ 적용대상\n○ 사무기기, 가전제품, 생활용품, 건축자재 등 156개 대상제품군\n□ 인증현황\n○ 2,737개 기업의 16,647개 제품(2015.12월말 기준)',
]
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)

python

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: 북한가족법 몇 차 개정에서 이혼판결 확정 후 3개월 내에 등록시에만 유효하다는 조항을 확실히 했을까?",
               "passage: 1990년에 제정된 북한 가족법은 지금까지 4차례 개정되어 현재에 이르고 있다. 1993년에 이루어진 제1차 개정은 주로 규정의 정확성을 기하기 위하여 몇몇 조문을 수정한 것이며, 실체적인 내용을 보완한 것은 상속의 승인과 포기기간을 설정한 제52조 정도라고 할 수 있다. 2004년에 이루어진 제2차에 개정에서는 제20조제3항을 신설하여 재판상 확정된 이혼판결을 3개월 내에 등록해야 이혼의 효력이 발생한다는 것을 명확하게 하였다. 2007년에 이루어진 제3차 개정에서는 부모와 자녀 관계 또한 신분등록기관에 등록한 때부터 법적 효력이 발생한다는 것을 신설(제25조제2항)하였다. 또한 미성년자, 노동능력 없는 자의 부양과 관련(제37조제2항)하여 기존에는 “부양능력이 있는 가정성원이 없을 경우에는 따로 사는 부모나 자녀, 조부모나 손자녀, 형제자매가 부양한다”고 규정하고 있었던 것을 “부양능력이 있는 가정성원이 없을 경우에는 따로 사는 부모나 자녀가 부양하며 그들이 없을 경우에는 조부모나 손자녀, 형제자매가 부양한다”로 개정하였다.",
               "passage: 환경마크 제도, 인증기준 변경으로 기업부담 줄인다\n환경마크 제도 소개\n□ 개요\n○ 동일 용도의 다른 제품에 비해 ‘제품의 환경성*’을 개선한 제품에 로고와 설명을 표시할 수 있도록하는 인증 제도\n※ 제품의 환경성 : 재료와 제품을 제조․소비 폐기하는 전과정에서 오염물질이나 온실가스 등을 배출하는 정도 및 자원과 에너지를 소비하는 정도 등 환경에 미치는 영향력의 정도(「환경기술 및 환경산업 지원법」제2조제5호)\n□ 법적근거\n○ 「환경기술 및 환경산업 지원법」제17조(환경표지의 인증)\n□ 관련 국제표준\n○ ISO 14024(제1유형 환경라벨링)\n□ 적용대상\n○ 사무기기, 가전제품, 생활용품, 건축자재 등 156개 대상제품군\n□ 인증현황\n○ 2,737개 기업의 16,647개 제품(2015.12월말 기준)"]

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

This evaluation references the KURE GitHub repository. (https://github.com/nlpai-lab/KURE)
We conducted an evaluation on all Korean Retrieval Benchmarks registered in MTEB.

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
exp-models/dragonkue-KoEn-E5-Tiny	37	0.687496	0.34735	0.7925	0.6143	0.75978	0.93018	0.86503	0.50333
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.

📈 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: 3
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

bibtex

@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

bibtex

@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

bibtex

@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}
}

KURE

bibtex

@misc{KURE,
  publisher = {Youngjoon Jang, Junyoung Son, Taemin Lee},
  year = {2024},
  url = {https://github.com/nlpai-lab/KURE}
}

Limitations

Long texts will be truncated to at most 512 tokens.

Acknowledgements

Special thanks to lemon-mint for their valuable contribution in optimizing and compressing this model.