Stella-base-en-v2 Open-source English Text Embedding Model - Free Implementation of Sentence Similarity and Feature Extraction

Stella Base En V2

Developed by infgrad

stella-base-en-v2 is an English text embedding model based on sentence-transformers, focusing on sentence similarity and feature extraction tasks.

Text Embedding

PyTorch

EnglishOpen Source License:MIT #High-precision Text Classification #Multi-task Text Embedding #English Semantic Similarity

Downloads 16.89k

Release Time : 10/19/2023

Model Overview

This model is primarily used to generate high-quality sentence embeddings, suitable for various natural language processing tasks such as sentence similarity calculation, text classification, and information retrieval.

Model Features

High-performance Sentence Embedding

Capable of generating high-quality sentence embeddings suitable for various downstream tasks.

Multi-task Support

Supports multiple natural language processing tasks such as sentence similarity, text classification, and information retrieval.

Comprehensive Evaluation

Extensively evaluated on multiple standard datasets with excellent performance.

Model Capabilities

Sentence similarity calculation

Text classification

Information retrieval

Clustering analysis

Reranking tasks

Use Cases

E-commerce

Product Review Classification

Classify Amazon product reviews to identify positive and negative feedback.

Achieved 93.26% accuracy on the AmazonPolarityClassification dataset.

Counterfactual Classification

Identify counterfactual statements in Amazon product reviews.

Achieved 77.19% accuracy on the AmazonCounterfactualClassification dataset.

Question Answering Systems

Duplicate Question Detection

Identify duplicate technical questions in the AskUbuntu community.

Achieved a MAP of 62.72 on the AskUbuntuDupQuestions dataset.

Academic Research

Paper Clustering

Perform clustering analysis on academic papers from arXiv and bioRxiv.

Achieved a V-measure of 47.24 on the ArxivClusteringP2P dataset.

🚀 stella-base-en-v2

This is a model related to sentence-transformers, which can be used for feature extraction, sentence similarity calculation, etc. It has been tested on multiple datasets in the MTEB benchmark, and the following are its performance metrics.

📚 Documentation

Model Information

Property	Details
Tags	sentence-transformers, feature-extraction, sentence-similarity, mteb
Model Name	stella-base-en-v2

Performance Metrics

1. Classification Tasks

Dataset	Accuracy	AP	F1
MTEB AmazonCounterfactualClassification (en)	77.19402985074628	40.43267503017359	71.15585210518594
MTEB AmazonPolarityClassification	93.256675	90.00824833079179	93.2473146151734
MTEB AmazonReviewsClassification (en)	49.612	-	48.530785631574304
MTEB Banking77Classification	84.72727272727273	-	84.29572512364581

2. Retrieval Tasks

Dataset	MAP@1	MAP@10	MAP@100	MAP@1000	MRR@1	MRR@10	MRR@100	MRR@1000	NDCG@1	NDCG@10	NDCG@100	NDCG@1000	Precision@1	Precision@10	Precision@100	Precision@1000	Recall@1	Recall@10	Recall@100	Recall@1000
MTEB ArguAna	37.411	52.673	53.410999999999994	53.415	37.838	52.844	53.581999999999994	53.586	37.411	60.626999999999995	63.675000000000004	63.776999999999994	37.411	8.578	0.989	0.1	37.411	85.775	98.86200000000001	99.644
MTEB CQADupstackAndroidRetrieval	32.822	44.139	45.786	45.906000000000006	41.059	50.751000000000005	51.548	51.583999999999996	41.059	50.573	56.25	58.004	41.059	9.757	1.609	0.20600000000000002	32.822	61.794000000000004	85.64699999999999	96.836
MTEB CQADupstackEnglishRetrieval	29.579	39.787	40.976	41.108	37.516	45.822	46.454	46.495999999999995	37.516	45.5	49.707	51.842	37.516	8.713	1.38	0.188	29.579	55.458	73.49799999999999	87.08200000000001
MTEB CQADupstackGamingRetrieval	40.489999999999995	53.313	54.290000000000006	54.346000000000004	46.27	56.660999999999994	57.274	57.301	46.27	59.192	63.026	64.079	46.27	9.511	1.23	0.136	40.489999999999995	73.148	89.723	97.073
MTEB CQADupstackGisRetrieval	26.197	35.135	36.14	36.216	28.475	37.096000000000004	38.006	38.06	28.475	40.263	45.327	47.225	28.475	6.249	0.919	0.11199999999999999	26.197	54.17999999999999	77.768	91.932
MTEB CQADupstackMathematicaRetrieval	16.683	25.013999999999996	26.411	26.531	20.896	29.758000000000003	30.895	30.964999999999996	20.896	30.294999999999998	36.68	39.519	20.896	5.56	1.006	0.13899999999999998	16.683	41.807	69.219	-

3. Clustering Tasks

Dataset	V-Measure
MTEB ArxivClusteringP2P	47.24219029437865
MTEB ArxivClusteringS2S	40.474604844291726
MTEB BiorxivClusteringP2P	39.55532460397536
MTEB BiorxivClusteringS2S	35.91195973591251

4. Reranking Task

Dataset	MAP	MRR
MTEB AskUbuntuDupQuestions	62.720542706366054	75.59633733456448

5. STS Task

Dataset	Cosine Similarity Pearson	Cosine Similarity Spearman	Euclidean Pearson	Euclidean Spearman	Manhattan Pearson	Manhattan Spearman
MTEB BIOSSES	86.31345008397868	85.94292212320399	85.03974302774525	85.88087251659051	84.91900996712951	85.96701905781116

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