Hi there,
Thank you for releasing such impressive work! I really appreciate the benchmark’s breadth and rigor, particularly its maze-based visual planning tasks. We have studied a closely related problem in our Amaze benchmark (Probing Visual Planning in Image Editing Models).
Perhaps Amaze could serve as a complementary evaluation in a future update of VBVR-Pro—we would be happy to help if useful!
Best,
Yann
Hi there,
Thank you for releasing such impressive work! I really appreciate the benchmark’s breadth and rigor, particularly its maze-based visual planning tasks. We have studied a closely related problem in our Amaze benchmark (Probing Visual Planning in Image Editing Models).
Perhaps Amaze could serve as a complementary evaluation in a future update of VBVR-Pro—we would be happy to help if useful!
Best,
Yann