search

mertyg / vision-language-models-are-bows

Experiments and data for the paper "When and why vision-language models behave like bags-of-words, and what to do about it?" Oral @ ICLR 2023
MIT License
222 stars 14 forks source link

parameter file problem #28

Closed haoshuai714 closed 1 year ago

haoshuai714 commented 1 year ago

Thank you so much for your excellent work. However, I have been unable to reproduce the results in the paper when trying to reproduce the NegCLIP code. Can you provide me a parameter file of your train stage? Thank you so much!

vinid commented 1 year ago

Hello and thanks!

Which results are you currently getting and what have you tried so far?

vinid commented 1 year ago

Closing this for now!