Semantic Textual Similarity On Sts16

평가 지표

Spearman Correlation

평가 결과

이 벤치마크에서 각 모델의 성능 결과

모델 이름	Spearman Correlation	Paper Title	Repository
PromptEOL+CSE+LLaMA-30B	0.8627	Scaling Sentence Embeddings with Large Language Models
DiffCSE-RoBERTa-base	0.8212	DiffCSE: Difference-based Contrastive Learning for Sentence Embeddings
Trans-Encoder-BERT-base-bi (unsup.)	0.8305	Trans-Encoder: Unsupervised sentence-pair modelling through self- and mutual-distillations
Trans-Encoder-RoBERTa-large-cross (unsup.)	0.8503	Trans-Encoder: Unsupervised sentence-pair modelling through self- and mutual-distillations
Trans-Encoder-BERT-large-bi (unsup.)	0.8481	Trans-Encoder: Unsupervised sentence-pair modelling through self- and mutual-distillations
BERTlarge-flow (target)	0.7763	On the Sentence Embeddings from Pre-trained Language Models	-
Trans-Encoder-RoBERTa-base-cross (unsup.)	0.8377	Trans-Encoder: Unsupervised sentence-pair modelling through self- and mutual-distillations
PromptEOL+CSE+OPT-13B	0.8590	Scaling Sentence Embeddings with Large Language Models
AnglE-LLaMA-13B	0.8700	AnglE-optimized Text Embeddings
PromCSE-RoBERTa-large (0.355B)	0.8496	Improved Universal Sentence Embeddings with Prompt-based Contrastive Learning and Energy-based Learning
IS-BERT-NLI	0.7016	An Unsupervised Sentence Embedding Method by Mutual Information Maximization
AnglE-LLaMA-7B	0.8691	AnglE-optimized Text Embeddings
Mirror-RoBERTa-base (unsup.)	0.78	Fast, Effective, and Self-Supervised: Transforming Masked Language Models into Universal Lexical and Sentence Encoders
DiffCSE-BERT-base	0.8054	DiffCSE: Difference-based Contrastive Learning for Sentence Embeddings
PromptEOL+CSE+OPT-2.7B	0.8591	Scaling Sentence Embeddings with Large Language Models
Dino (STSb/̄	0.7718	Generating Datasets with Pretrained Language Models
SimCSE-RoBERTalarge	0.8393	SimCSE: Simple Contrastive Learning of Sentence Embeddings
AnglE-LLaMA-7B-v2	0.8700	AnglE-optimized Text Embeddings
SRoBERTa-NLI-large	0.7682	Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks
Mirror-BERT-base (unsup.)	0.743	Fast, Effective, and Self-Supervised: Transforming Masked Language Models into Universal Lexical and Sentence Encoders

0 of 20 row(s) selected.