SynCo: Synthetic Hard Negatives in Contrastive Learning for Better Unsupervised Visual Representations

Open paper
Benchmark Model Rank Results
self-supervised-image-classification-on-imagenetSynCo (ResNet-50) 800ep#95Top 1 Accuracy: 70.6%Top 5 Accuracy: 89.8%Number of Params: 24M
self-supervised-image-classification-on-imagenetSynCo (ResNet-50) 200ep#101Top 1 Accuracy: 67.9%Top 5 Accuracy: 88Number of Params: 24M
semi-supervised-image-classification-on-imagenet-1-labeled-dataSynCo (ResNet-50) 800ep#45Top 1 Accuracy: 50.8%Top 5 Accuracy: 77.5%Number of params: 24M
semi-supervised-image-classification-on-imagenet-10-labeled-dataSynCo (ResNet-50) 800ep#39Top 1 Accuracy: 66.6%Top 5 Accuracy: 88.0%Number of params: 24M

Rank counts only results with a code link.