Closed TheShadow29 closed 4 years ago
Update: Seems like there is a mis-match between the full dataset
and modeling quickstart
provide here: https://github.com/askforalfred/alfred/tree/master/data
It works with modeling quickstart
data.
Thanks @TheShadow29! There might be some trajectories left over from the generation phase without language annotations. Will check it out soon...
Fixed in the latest full_2.1.0.zip. Data Readme has been updated.
Hi
I am trying to run the preprocessing steps, but facing an error in
process_language
(https://github.com/askforalfred/alfred/blob/master/data/preprocess.py#L93)In
train
andvalid
splits, it seemsridx >= len(ex['turk_annotations']['anns'])
for quite a few annotations. This causes the error in L93.I am likely missing something here. Any clue as to what is going on? Thank you for your patience.