Conversation
|
MLCommons CLA bot All contributors have signed the MLCommons CLA ✍️ ✅ |
|
Right, thanks! Can you sign the MLCommons Association CLA first? See https://mlcommons.org/community/subscribe/ |
|
@ccl-core Working on it, LBNL/NERSC ( my employer ) has signed it, just trying to get added to it. We emailed support@mlcommons.org, it that the right place for this request? |
|
recheck |
|
@ccl-core Should be good now. |
README.md
Outdated
| builder.download_and_prepare() | ||
| # 4. Split for training/testing | ||
| train, test = builder.as_data_source(split=['default[:80%]', 'default[80%:]']) | ||
| train, test = builder.as_data_source(split=['train[:80%]', 'test[80%:]']) |
There was a problem hiding this comment.
the 80% split was, IIRC, just to make up for the missing split definiton in the early croissant definitions for huggingface dataset. Can't you just train, test = builder.as_data_source(split=['train', 'test']) ?
|
Looks good, thanks! Can you please fix the |
Done |
Looks like the record set name has changed also with the split names.
2 participants
Looks like the record set name has changed also with the split names.