create_tiny_dataset.py

from datasets import load_dataset
from datasets import Dataset

# 1. Load the en split of C4 with streaming
ds = load_dataset("allenai/c4", "en", split="train", streaming=True)

# 2. Take first 10k examples and convert to regular dataset
ds_small = list(ds.take(10000))  # take first 10k examples
ds_small = Dataset.from_list(ds_small)  # convert to regular Dataset
assert len(ds_small) == 10000
print("Sample example:", ds_small[0])

# 4. Push to the Hub
repo_name = "boom-project/c4-en-10k-debug"

ds_small.push_to_hub(
    repo_id=repo_name,
    token=True,       # uses the token from `huggingface-cli login`
    private=False,    
)

print(f"✅ Pushed 10,000-example C4 subset to https://huggingface.co/{repo_name}")

# After pushing to the Hub, verify the dataset size and inspect one example
# from datasets import load_dataset
# ds = load_dataset("boom-project/c4-en-10k-debug", split="train")
# assert len(ds) == 10000
# print("Sample example:", ds[0])