Datasets:

Modalities:
Tabular
Text
Formats:
parquet
Languages:
code
ArXiv:
Libraries:
Datasets
Dask
License:

Incredible work! Question on the dataset format

#2
by yuxiang630 - opened
BigCode org

Thanks for the incredible work on StarCoder2 and the great contributions to open science!! I was trying to study with the dataset right away but just got a bit confused with the statement "We are working on making the training datasets available in the coming weeks". I suppose this dataset is the training set for 3B/7B models except that it only contains IDs. Does the statement mean in the upcoming weeks the contents can also be downloaded directly from the hub? If so, are the contents the same as retrieving using the IDs?

BigCode org

Hi @yuxiang630 ! The upcoming training datasets will have all of the file contents included (with redacted PII).

Sign up or log in to comment