Fetching the paper…

Text Descriptions are Compressive and Invariant Representations for Visual Learning · Around