Datasets:

Languages:
English
ArXiv:
License:

Not able to load the dataset

#10
by surya-narayanan - opened

I'm not able to load the dataset. Here is the code I'm using

dataset_name = "EleutherAI/the_pile"
dataset = load_dataset(dataset_name, subsets = ['hacker_news', 'enron_emails'])

Here is the rror I'm getting

TypeError: Couldn't cast array of type
struct<file: string, id: string>
to
{'id': Value(dtype='string', id=None)}

Any ideas why?

Thanks for reporting, @surya-narayanan .

Yes, I think there is an issue here. Before, it used to work because there was some kind of implicit cast of type dict to str. However, this implicit cast no longer works and the script should be fixed to perform this cast explicitly.

mariosasko changed discussion status to closed
surya-narayanan changed discussion status to open

Hey, still facing this issue- when I try to download the github and pubmed datasets

struct<file_name: string, language: string, mime_type: string, pmid: int64, repo_language: string, repo_name: string, stars: string>
to
{'language': Value(dtype='string', id=None), 'pmid': Value(dtype='int64', id=None)}

Sign up or log in to comment