Datasets:

Modalities:
Audio
Text
Formats:
parquet
ArXiv:
Libraries:
Datasets
Dask

Editing SLUE-TED data and releasing test labels

#9
by roshansh - opened

Changes in this PR:

  1. Added test labels to enable evaluation.
  2. Removed one test sample with only music -> impossible to make predictions and should not be scored.
  3. Updated the text pre-processing on abstract and titles for train,dev and test splits so as to remove non-standard text characters.
  4. Renamed the fine-tune set to train as per common convention.

Hi Roshan. I don't see any update on this pr. Can you check?

I seem to be having trouble with uploading the large zip files. Will try to resolve soon.

Added all files - please check and merge.

roshansh changed pull request status to open
sshon-asapp changed pull request status to merged

Sign up or log in to comment