ca0e937
1
[MMCOMPOSITION: Revisiting the Compositionality of Pre-trained Vision-Language Models](https://arxiv.org/pdf/2410.09733): https://arxiv.org/pdf/2410.09733
