Skip to content

Conversation

MaoSong2022
Copy link
Collaborator

fix argument error in MMMU_Pro benchmark, close #929

@kennymckormick
Copy link
Member

Hi, @MaoSong2022 ,

It looks like every other dataset do not set the arg names explicitly, so I don't think the modification is necessary.

@MaoSong2022
Copy link
Collaborator Author

the change is to keep the arguments consistent with ImageBaseDataset initialization. For now, only a few dataset implements this method.

  • ImageBaseDataset initialization: def __init__(self, dataset='MMBench', skip_noimg=True)
  • some other dataset initialization (such as MMGenBench): def __init__(self, dataset, **kwargs)
  • this commit (related issue): def __init__(self, dataset='MMMU_Pro', skip_noimg=True)

the question is should we use consistent arguments when inheriting ImageBaseDataset, if the answer is yes, which format should we use.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[bug] 反馈两个bug,模型和数据集用config.json配置会出现一些问题

2 participants