Dataset scripts are no longer supported

I was using earlier GeneratorBasedBuilder class for loading database now i am getting below error :
Exception occurred: Dataset scripts are no longer supported.

using load_dataset for loading script . but somehow it is not supported not

plz tell me what is the other way to load database using GeneratorBasedBuilder class i need to pre process the database before saving in arrow or other format.

2 Likes

Seems trust_remote_code is deprecated in datasets 4.0.0.
So quick workarounds:

pip install datasets<4.0.0 

In addition, it seems that downgrading huggingface_hub may be necessary in some cases.

1 Like

I clarify this just in case. It seems that support for the function to build datasets locally would continue.

yes, we can not use load_dataset if implementing a Builder class.
so need to explicitly call builder class and generate dataset.
builder.download_and_prepare()
dataset = builder.as_dataset(split=Split.TRAIN)

1 Like

This topic was automatically closed 12 hours after the last reply. New replies are no longer allowed.