multilingual-e5-small: An open-source multilingual text processing model - Supports diverse tasks such as classification and retrieval.

Sentence Transformers Multilingual E5 Small

Developed by beademiguelperez

multilingual - e5 - small is a model that performs excellently in multilingual text processing tasks, supporting various tasks such as classification, retrieval, clustering, re - ranking, and semantic text similarity.

Text Embedding Supports Multiple LanguagesOpen Source License:MIT #Multilingual text embedding #Cross - language retrieval #Semantic similarity calculation

Downloads 3,922

Release Time : 3/25/2024

Model Overview

This model performs outstandingly in multilingual text processing tasks and is suitable for a variety of natural language processing scenarios, including but not limited to text classification, information retrieval, and cluster analysis.

Model Features

Multilingual support

Supports text processing in multiple languages, including English, German, Spanish, French, Japanese, and Chinese, etc.

Multi - functional task support

Suitable for a variety of natural language processing tasks, such as classification, retrieval, clustering, re - ranking, and semantic text similarity.

High performance

Performs excellently on multiple datasets, especially in classification and retrieval tasks.

Model Capabilities

Text classification

Information retrieval

Cluster analysis

Re - ranking

Semantic text similarity calculation

Bilingual mining

Use Cases

E - commerce

Amazon review classification

Perform sentiment classification on Amazon product reviews, supporting multiple languages.

The accuracy rate reaches 44.698% in English review classification and 40.246% in German review classification.

Finance

Bank customer service question classification

Automatically classify bank customer service questions to improve customer service efficiency.

The accuracy rate reaches 79.416% on the Banking77 dataset.

Information retrieval

Question - answering system

Used for information retrieval and re - ranking in question - answering systems.

The MAP@10 reaches 56.168% on the HotpotQA dataset.

🚀 multilingual-e5-small

This document presents the performance metrics of the multilingual-e5-small model on various natural language processing tasks, including classification, retrieval, clustering, reranking, STS, and bitext mining.

📚 Documentation

Model Information

Property	Details
Model Name	multilingual-e5-small
Tags	mteb, Sentence Transformers, sentence-similarity, sentence-transformers

Task Performance Metrics

Classification Tasks

Dataset	Language	Accuracy	AP	F1
MTEB AmazonCounterfactualClassification	en	73.79104477611939	36.9996434842022	67.95453679103099
MTEB AmazonCounterfactualClassification	de	71.64882226980728	82.11942130026586	69.87963421606715
MTEB AmazonCounterfactualClassification	en-ext	75.8095952023988	24.46869495579561	63.00108480037597
MTEB AmazonCounterfactualClassification	ja	64.186295503212	15.496804690197042	52.07153895475031
MTEB AmazonPolarityClassification	default	88.699325	85.27039559917269	88.65556295032513
MTEB AmazonReviewsClassification	en	44.69799999999999	-	43.73187348654165
MTEB AmazonReviewsClassification	de	40.245999999999995	-	39.3863530637684
MTEB AmazonReviewsClassification	es	40.394	-	39.301223469483446
MTEB AmazonReviewsClassification	fr	38.864	-	37.97974261868003
MTEB AmazonReviewsClassification	ja	37.682	-	37.07399369768313
MTEB AmazonReviewsClassification	zh	37.504	-	36.62317273874278
MTEB Banking77Classification	default	79.41558441558442	-	79.25886487487219
MTEB EmotionClassification	default	42.455	-	37.59462649781862
MTEB ImdbClassification	default	80.	-	-

Retrieval Tasks

Dataset	MAP@1	MAP@10	MAP@100	MAP@1000	MAP@3	MAP@5	MRR@1	MRR@10	MRR@100	MRR@1000	MRR@3	MRR@5	NDCG@1	NDCG@10	NDCG@100	NDCG@1000	NDCG@3	NDCG@5	Precision@1	Precision@10	Precision@100	Precision@1000	Precision@3	Precision@5	Recall@1	Recall@10	Recall@100	Recall@1000	Recall@3	Recall@5
MTEB ArguAna	19.061	31.703	32.967	33.001000000000005	27.466	29.564	19.559	31.874999999999996	33.146	33.18	27.667	29.74	19.061	39.062999999999995	45.184000000000005	46.115	30.203000000000003	33.953	19.061	6.279999999999999	0.9129999999999999	0.099	12.706999999999999	9.431000000000001	19.061	62.802	91.323	98.72	38.122	47.155
MTEB CQADupstackRetrieval	24.252999999999997	31.655916666666666	32.680749999999996	32.79483333333334	29.43691666666666	30.717416666666665	28.602750000000004	35.56875	36.3595	36.427749999999996	33.586166666666664	34.73641666666666	28.602750000000004	36.06933333333334	40.70141666666667	43.24341666666667	32.307916666666664	34.129999999999995	28.602750000000004	6.097666666666667	0.9809166666666668	0.13766666666666663	14.628166666666667	10.266916666666667	24.252999999999997	45.31916666666667	66.03575000000001	83.94708333333334	34.71941666666666	39.46358333333333
MTEB ClimateFEVER	9.024000000000001	15.644	17.154	17.345	13.028	14.251	19.674	29.826999999999998	30.935000000000002	30.987	26.645000000000003	28.29	19.674	22.545	29.207	32.912	17.952	19.363	19.674	7.212000000000001	1.435	0.212	13.507	10.397	9.024000000000001	28.077999999999996	51.403	72.406	16.768	20.737
MTEB DBPedia	8.012	17.138	24.146	25.622	12.552	14.435	62.25000000000001	71.186	71.504	71.514	69.333	70.408	49.75	37.76	42.071	49.309	41.644	39.812999999999995	62.25000000000001	30.15	9.753	1.9189999999999998	45.667	39.15	8.012	22.599	48.068	71.328	14.043	17.124
MTEB FEVER	58.092	69.586	69.968	69.982	67.48100000000001	68.915	62.166	73.588	73.86399999999999	73.868	71.6	72.99	62.166	75.27199999999999	76.816	77.09700000000001	71.36	73.785	62.166	9.716	1.065	0.11	28.278	18.343999999999998	58.092	88.73400000000001	95.195	97.04599999999999	78.45	84.316
MTEB FiQA2018	16.649	26.457000000000004	28.169	28.352	23.305	25.169000000000004	32.407000000000004	40.922	41.931000000000004	41.983	38.786	40.205999999999996	32.407000000000004	33.314	40.312	43.685	30.391000000000002	31.525	32.407000000000004	8.966000000000001	1.6019999999999999	0.22200000000000003	20.165	14.722	16.649	39.117000000000004	65.726	85.784	27.914	33.289
MTEB HotpotQA	36.253	56.16799999999999	57.06099999999999	57.126	52.644999999999996	54.909	72.505	79.66	79.869	79.88	78.411	79.19800000000001	72.505	65.094	68.219	69.515	59.99	62.909000000000006	72.505	13.749	1.619	0.179	38.357	25.313000000000002	36.253	68.744	80.925	89.534	57.535000000000004	63.282000000000004

Clustering Tasks

Dataset	V-Measure
MTEB ArxivClusteringP2P	39.22266660528253
MTEB ArxivClusteringS2S	30.79980849482483
MTEB BiorxivClusteringP2P	35.747820820329736
MTEB BiorxivClusteringS2S	27.045143830596146

Reranking Task

Dataset	MAP	MRR
MTEB AskUbuntuDupQuestions	57.8790068352054	71.78791276436706

STS Task

Dataset	Cosine Similarity Pearson	Cosine Similarity Spearman	Euclidean Pearson	Euclidean Spearman	Manhattan Pearson	Manhattan Spearman
MTEB BIOSSES	82.36328364043163	82.26211536195868	80.3183865039173	79.88495276296132	80.14484480692127	80.39279565980743

Bitext Mining Tasks

Dataset	Language Pair	Accuracy	F1	Precision	Recall
MTEB BUCC	de-en	98.0375782881002	97.86012526096033	97.77139874739039	98.0375782881002
MTEB BUCC	fr-en	93.35241030156286	92.66050333846944	92.3306919069631	93.35241030156286
MTEB BUCC	ru-en	94.0699688257707	93.50236693222492	93.22791825424315	94.0699688257707
MTEB BUCC	zh-en	89.25750394944708	88.79234684921889	88.57293312269616	89.25750394944708

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご