Also, the paper is only on arxiv, we have no plans to submit, and the code is super simple
If anyone wants to add results we're flexible with it, and we can add new authors
Our goal is to have a reference paper for anyone doing retrieval, so we can update the paper as needed
Curious about image retrieval and contrastive learning?
We release the most comprehensive retrieval benchmark, with thousands of experiments on 4 diverse datasets, dozens of losses, investigating batch sizes, LRs, training techniques, data labeling, and much more!
A summary ⬇️