oxai / debias-vision-lang Goto Github PK

Getting the following error when trying to import the library:
ModuleNotFoundError Traceback (most recent call last)
in <cell line: 1>()
----> 1 import debias_clip

/usr/local/lib/python3.10/dist-packages/debias_clip/init.py in
24
25 from .model import *
---> 26 from measuring_bias import measure_bias

ModuleNotFoundError: No module named 'measuring_bias'

You provide code for measuring NDKL and MaxSkew where the desired distribution for the similar images is based on either demographic parity or equality of opportunity. When reporting results in your paper, which of these did you use? Apologies if this is somewhere I missed!

Hi,

I want to reproduce the debiasing training using my own prompts. Do you plan to publish the training script, including the hyperparameters used?

Thank you.

Thanks a lot for making your code public. It's very useful.

I found a couple of mistakes, 1) I had to change "from measuring_bias" to "from .measuring_bias" otherwise I was getting an import error. 2) I had to move debias_tokens to gpu explicitly with self.debias_tokens = self.debias_tokens.to(text.device)

Now, I am getting the following error in encode_text function:

text_features = self.clip.transformer(text_features)

RuntimeError: mat1 and mat2 must have the same dtype

I have tried to change the dtype variable (model.py line 168 to float32, float64), I have also tried to explicitly set the text_features to self.dtype before calling clip.transformer (model.py line 257) but no dice. Could you perhaps help?