M

Multilingual E5 Small Ko V2

Developed by dragonkue
A Korean sentence transformer fine-tuned based on intfloat/multilingual-e5-small for Korean retrieval tasks
Downloads 252
Release Time : 6/10/2025

Model Overview

This is a model based on sentence-transformers, specifically optimized for Korean retrieval tasks. It can map text to a 384-dimensional vector space and is suitable for various tasks such as semantic similarity calculation and semantic search.

Model Features

Lightweight design
Suitable for running demos or lightweight applications, achieving a good balance between speed and accuracy
Excellent performance
In Korean benchmark tests, the small-sized model outperforms the larger 'intfloat/multilingual-e5-base' model
Scalability
Can be used in combination with a re-ranker to further improve retrieval performance
Model fusion technology
By weighted averaging to merge the Korean-specific model and the basic multilingual model, optimal performance is obtained

Model Capabilities

Semantic text similarity calculation
Semantic search
Paraphrase mining
Text classification
Text clustering

Use Cases

Information retrieval
Korean document retrieval
Retrieve relevant documents from a Korean document library
Performs excellently in multiple Korean retrieval benchmark tests
Open-domain question answering
Used for paragraph retrieval in Korean open-domain question answering systems
Performs well on the Ko-StrategyQA dataset
Text analysis
Semantic similarity calculation
Calculate the semantic similarity between two Korean sentences
Text clustering
Cluster Korean texts according to semantic similarity
Featured Recommended AI Models
ยฉ 2025AIbase