search

mlfoundations / wise-ft

Robust fine-tuning of zero-shot models
https://arxiv.org/abs/2109.01903
Other
654 stars 67 forks source link

Modify for object detection #4

Open nikky4D opened 2 years ago

nikky4D commented 2 years ago

Is it possible to use this with a multilabel dataset? with up to 5 objects per image?

mitchellnw commented 2 years ago

Yes this will probably still work, though we don't have code currently for this.

vkrishnamurthy11 commented 1 year ago

I tried using this for a multi label dataset where each input can belong to one or more of two categories. I get the following error: RuntimeError: stack expects each tensor to be equal size, but got [1] at entry 0 and [2] at entry 4

I think some code change will be required if we want to support this functionality.