metadata
dataset_info:
features:
- name: text
dtype: string
- name: concept_with_offset
dtype: string
splits:
- name: train
num_bytes: 75286835
num_examples: 6160
download_size: 38908375
dataset_size: 75286835
Dataset Card for "bookcorpus_compact_1024_test"
6160 samples randomly sampled from the shard9 of Bookcorpus_compact_1024
from datasets import load_dataset
from datasets import Dataset
corpus_name="xxx"
ds = load_dataset(corpus_name, split="train")
shuffled_ds = ds.shuffle(seed=42)
test_ds = Dataset.from_dict{shuffled_ds[:6160]} # len(ds)//10