train_scripts/sample_preview.py

import random
from datasets import load_dataset

def main():
    ds = load_dataset("json", data_files={"train":"data/raw/train_small.jsonl"}, split="train")
    print("Columns:", ds.column_names)
    for _ in range(5):
        ex = ds[random.randint(0, len(ds)-1)]
        print("\n---")
        print("id:", ex.get("id"))
        print("text:", (ex.get("text") or "")[:300])

if __name__ == "__main__":
    main()