: Vocabulaire du TAL: contrastive loss

NLP resources and evaluation > measure > training loss > contrastive loss

contrastive loss

A loss function used in training neural network models which draws representations of similar inputs close to one another, while simultaneously pushing representations of different inputs apart, and is used for tasks such as sentence similarity, semantic textual similarity, and sentence embedding learning. (Based on Moayeri et al., A Comprehensive Study of Image Classification Model Sensitivity to Foregrounds, Backgrounds, and Visual Attributes, 2022)

http://data.loterre.fr/ark:/67375/8LP-ZBB710P2-5

RDF/XML TURTLE JSON-LD Dernière modification le 21/05/2024

Loterre