e-bug / volta

[TACL 2021] Code and data for the framework in "Multimodal Pretraining Unmasked: A Meta-Analysis and a Unified Framework of Vision-and-Language BERTs"
https://aclanthology.org/2021.tacl-1.58/
MIT License
113 stars 24 forks source link

Generating hard negative captions in Flick30k dataset #4

Closed LandyGuo closed 3 years ago

LandyGuo commented 3 years ago

A download link of hard negative captions is provided for the Flick30k dataset here. But neither explanations nor scripts are found for generating those hard negatives.
Could you help reproduce the generation process? @e-bug @elliottd

e-bug commented 3 years ago

We have re-used the ones uploaded for ViLBERT-MT. I believe this script generates them. Let us know if they match!

LandyGuo commented 3 years ago

thanks. Verified that they are exactly match. closing it now