Issue with generating lmdb

facebookresearch / vilbert-multi-task

Multi Task Vision and Language

MIT License

797 stars 180 forks source link

Issue with generating lmdb #11

Open Princesston opened 4 years ago

Princesston commented 4 years ago

Running the script/convert_to_lmdb.py stops at around 90% with no error, but apparently it hasn't finished, as running train_tasks.py gives TypeError: a bytes-like object is required, not 'NoneType' at here. Does anyone know what could be the problem? Thanks~

jiasenlu commented 4 years ago

yeah, it will have a problem. Since we encode the mapping dict in the last line of covert_to_lmdb.py script. Could you try the script again? or convert with fewer images?

vedanuj commented 4 years ago

All the data for the downstream tasks can be downloaded now: Please check this update #26

Readme here : https://github.com/facebookresearch/vilbert-multi-task/tree/master/data