in pyspark_huggingface/huggingface_sink.py [0:0]
def __init__(self, options):
super().__init__(options)
if "path" not in options or not options["path"]:
raise Exception("You must specify a dataset name.")
from huggingface_hub import get_token
kwargs = dict(self.options)
self.repo_id = kwargs.pop("path")
self.path_in_repo = kwargs.pop("path_in_repo", None)
self.split = kwargs.pop("split", None)
self.revision = kwargs.pop("revision", None)
self.token = kwargs.pop("token", None) or get_token()
self.endpoint = kwargs.pop("endpoint", None)
for arg in kwargs:
if kwargs[arg].lower() == "true":
kwargs[arg] = True
elif kwargs[arg].lower() == "false":
kwargs[arg] = False
else:
try:
kwargs[arg] = ast.literal_eval(kwargs[arg])
except ValueError:
pass
self.kwargs = kwargs