Datasets:
Tasks:
Text Retrieval
Sub-tasks:
document-retrieval
Languages:
English
Tags:
text-retrieval
License:
Can't load dataset
#1
by
Samoed
- opened
Hi! I tried to download dataset with datasets==2.20.0
and ==3.0.1
and they both raised error
datasets.exceptions.NonMatchingSplitsSizesError: [{'expected': SplitInfo(name='test', num_bytes=130413, num_examples=2850, shard_lengths=None, dataset_name=None), 'recorded': SplitInfo(name='test', num_bytes=161729, num_examples=2849, shard_lengths=None, dataset_name='webis-touche2020-v3')}]
Code:
from datasets import load_dataset
ds = load_dataset("castorini/webis-touche2020-v3")
only this work
from datasets import VerificationMode, load_dataset
ds = load_dataset("castorini/webis-touche2020-v3", verification_mode=VerificationMode.NO_CHECKS)
but when I tried to download queries
repo = "castorini/webis-touche2020-v3"
qrels = load_dataset(repo, verification_mode=VerificationMode.NO_CHECKS)
corpus = load_dataset(repo, "corpus", verification_mode=VerificationMode.NO_CHECKS)
qeries = load_dataset(repo, "queries", verification_mode=VerificationMode.NO_CHECKS)
This error raised
DatasetGenerationCastError: An error occurred while generating the dataset
All the data files must have the same columns, but at some point there are 1 new columns ({'metadata'})
This happened while the json dataset builder was generating data using
hf://datasets/castorini/webis-touche2020-v3/queries.jsonl (at revision 40a2b6e821b7762671b231561fba1088f5d3ab34)
Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)