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
114 stars 24 forks source link

LMDB file download #2

Closed darthgera123 closed 3 years ago

darthgera123 commented 3 years ago

Currently to download LMDB files , the link opens up the lmdb file and we download each content separately. Following this link. Can you share a link where we can directly download the lmdb file as we arent able to compress the contents back into lmdb format? @elliottd @e-bug

e-bug commented 3 years ago

Hi, I'm not sure I follow what you mean. As shown here, you just need to specify a path to the folder ending in .lmdb. So "refcoco+_unc_dets36_feat.lmdb" is not actually a file but a folder. Does this help?

darthgera123 commented 3 years ago

Thanks for replying. I thought lmdb was a compressed file like zip. Thanks for clearing that up. Also another question, do you have a smaller version of resnet101_faster_rcnn_genome_imgfeats/ for vqa which is 1TB, which as you can imagine is quite unfeasible for downloading as well as storing.

e-bug commented 3 years ago

I don't but you can preprocess a subset of the data by following the general instructions.