Open
Description
Hi,
Thanks for the very interesting work. I am just wondering if is there a script that will enable us to reproduce the results on V* and HRBench4K. It seems like the evaluation in the VLMEvalKit and the code for the DC2 are independent. What are the hyper-parameters like min_size, max_deep, alpha and theta to run experiments on these two datasets?
Thanks :)
Best,
Liangyu
Metadata
Metadata
Assignees
Labels
No labels