license: apache-2.0
Here we estimate the performance of the CLIP embeddings (contrastive training between text - image data).
Space using Forbu14/openai_clip_embeddings 1
Evaluation results
- accuracy on MTEB AmazonCounterfactualClassification (en)test set self-reported57.493
- ap on MTEB AmazonCounterfactualClassification (en)test set self-reported23.594
- f1 on MTEB AmazonCounterfactualClassification (en)test set self-reported52.202
- accuracy on MTEB AmazonReviewsClassification (en)test set self-reported30.590
- f1 on MTEB AmazonReviewsClassification (en)test set self-reported30.418
- accuracy on MTEB Banking77Classificationtest set self-reported73.419
- f1 on MTEB Banking77Classificationtest set self-reported73.365
- v_measure on MTEB BiorxivClusteringP2Ptest set self-reported29.331
- v_measure on MTEB BiorxivClusteringS2Stest set self-reported27.812
- accuracy on MTEB EmotionClassificationtest set self-reported33.620