Benchmark and analysis of 165 pretrained SSL models. Code for "Evaluating Self-Supervised Learning via Risk Decomposition". - View it on GitHub