Fetching the paper…

Iterated Learning Improves Compositionality in Large Vision-Language Models · Around