def get_dataset_names():
if is_local_mode() is True:
# fetch all dirs in "./datasets/"
datasets = [dataset_dir.split("/")[-2] for dataset_dir in glob.glob("./datasets/*/")]
else:
# fetch all dataset names
api = hf_api.HfApi()
datasets = [x.datasetId for x in api.dataset_list()]
dataset_names_parametrized = [{"testcase_name": x, "dataset_name": x} for x in datasets]
return dataset_names_parametrized
def get_dataset_names():
datasets = []
if is_local_mode() is True or isinstance(is_local_mode(),
int) and is_local_mode() > 0:
# fetch all dirs in "./datasets/"
datasets = datasets + [
dataset_dir.split("/")[-2]
for dataset_dir in glob.glob("./datasets/*/")
]
if is_aws_mode() is True or isinstance(is_aws_mode(),
int) and is_aws_mode() > 0:
api = hf_api.HfApi()
# fetch all dataset names
datasets = datasets + [x.id for x in api.dataset_list()]
dataset_names_parametrized = [{
"testcase_name": x,
"dataset_name": x
} for x in set(datasets)]
return dataset_names_parametrized
def get_aws_dataset_names():
api = hf_api.HfApi()
# fetch all dataset names
datasets = [x.id for x in api.dataset_list(with_community_datasets=False)]
return [{"testcase_name": x, "dataset_name": x} for x in datasets]
def get_aws_dataset_names():
api = hf_api.HfApi()
# fetch all dataset names
datasets = api.dataset_list(with_community_datasets=False, id_only=True)
return [{"testcase_name": x, "dataset_name": x} for x in datasets]
def get_aws_metric_names():
api = hf_api.HfApi()
# fetch all metric names
metrics = [x.id for x in api.metric_list()]
return [{"testcase_name": x, "metric_name": x} for x in metrics]