SGPT-1.3B Open-Source Sentence Transformer Model - Freely Achieve Sentence Similarity and Feature Extraction

SGPT 1.3B Weightedmean Msmarco Specb Bitfit

Developed by Muennighoff

SGPT-1.3B is a sentence transformer model optimized with weighted mean and bitfit techniques, primarily used for sentence similarity and feature extraction tasks.

Text Embedding

PyTorch

Downloads 87

Release Time : 3/2/2022

Model Overview

This model is a 1.3B-parameter sentence transformer, optimized with weighted mean and bitfit techniques, demonstrating strong performance across multiple text similarity and classification tasks.

Model Features

Weighted Mean Technique

Uses weighted mean method to optimize sentence representations, enhancing semantic understanding capabilities.

Bitfit Optimization

Employs bitfit technology for model fine-tuning, improving performance while maintaining efficiency.

Multi-task Support

Supports various natural language processing tasks, including classification, clustering, and retrieval.

Model Capabilities

Sentence Similarity Calculation

Text Feature Extraction

Text Classification

Information Retrieval

Text Clustering

Question Answering System Support

Use Cases

E-commerce

Product Review Classification

Sentiment polarity classification for Amazon product reviews

Achieved 73.21% accuracy on the MTEB Amazon Polarity Classification task

Counterfactual Analysis

Identifying counterfactual statements in Amazon product reviews

Achieved 65.21% accuracy on the MTEB Amazon Counterfactual Classification task

Academic Research

Paper Clustering

Topic clustering for arXiv academic papers

Achieved 43.38 V-measure on the MTEB Arxiv Clustering P2P task

Technical Support

Question Answering Retrieval

Retrieving related questions in technical forums

Achieved 28.48 average precision@10 on the MTEB CQADupstack Retrieval task

🚀 SGPT-1.3B-weightedmean-msmarco-specb-bitfit

This is a model related to sentence - transformers, which can be used for feature extraction, sentence similarity calculation, and has been tested on the MTEB benchmark.

📚 Documentation

Model Information

Property	Details
Model Type	SGPT-1.3B-weightedmean-msmarco-specb-bitfit
Tags	sentence-transformers, feature-extraction, sentence-similarity, mteb

Performance Metrics

The model has been evaluated on multiple tasks and datasets. Here are the detailed results:

Classification Tasks

Dataset	Accuracy	AP	F1
MTEB AmazonCounterfactualClassification (en)	65.20895522388061	29.59212705444778	59.97099864321921
MTEB AmazonPolarityClassification	73.20565	67.36680643550963	72.90420520325125
MTEB AmazonReviewsClassification (en)	34.955999999999996	N/A	34.719324437696955
MTEB Banking77Classification	82.05844155844156	N/A	82.0185837884764

Retrieval Tasks

The retrieval tasks are evaluated using multiple metrics such as map_at_*, mrr_at_*, ndcg_at_*, precision_at_*, and recall_at_*. Here are some examples:

Dataset	map_at_1	map_at_10	map_at_100
MTEB ArguAna	26.101999999999997	40.958	42.033
MTEB CQADupstackAndroidRetrieval	26.519	35.634	36.961
MTEB CQADupstackEnglishRetrieval	25.356	34.596	35.714
MTEB CQADupstackGamingRetrieval	32.759	43.937	45.004
MTEB CQADupstackGisRetrieval	18.962	25.863000000000003	26.817999999999998
MTEB CQADupstackMathematicaRetrieval	11.24	18.503	19.553

Clustering Tasks

Dataset	V - Measure
MTEB ArxivClusteringP2P	43.384194916953774
MTEB ArxivClusteringS2S	33.70962633433912
MTEB BiorxivClusteringP2P	35.05918333141837
MTEB BiorxivClusteringS2S	30.71055028830579

Reranking Task

Dataset	MAP	MRR
MTEB AskUbuntuDupQuestions	58.133058996870076	72.10922041946972

STS Task

Dataset	Cosine Similarity Pearson	Cosine Similarity Spearman	Euclidean Pearson	Euclidean Spearman	Manhattan Pearson	Manhattan Spearman
MTEB BIOSSES	86.62153841660047	83.01514456843276	86.00431518427241	83.85552516285783	85.83025803351181	83.86636878343106

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