You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
For the transfer attack implemented in the script LAVIS_tool/_train_adv_img_blip.py, the target image features come from the function where blip_model is loaded through lavis package. However, it seems like there is no implementation of the forward_encoder_image function in the source code of lavis, thus triggering the error: AttributeError: 'Blip2OPT' object has no attribute 'forward_encoder_image'. How to solve this issue?
The text was updated successfully, but these errors were encountered:
For the transfer attack implemented in the script LAVIS_tool/_train_adv_img_blip.py, the target image features come from the function where blip_model is loaded through lavis package. However, it seems like there is no implementation of the forward_encoder_image function in the source code of lavis, thus triggering the error: AttributeError: 'Blip2OPT' object has no attribute 'forward_encoder_image'. How to solve this issue?
The text was updated successfully, but these errors were encountered: