Provable Stochastic Optimization for Global Contrastive Learning: Small Batch Does Not Harm Performance

Open in new window