from huggingface_hub import snapshot_download
snapshot_download(
  repo_id="google/WaxalNLP", repo_type="dataset",
  local_dir="/scratch/data/waxal",
  allow_patterns=[
    "data/ASR/lin/lin-train-*.parquet","data/ASR/lin/lin-validation-*.parquet",
    "data/ASR/sna/sna-train-*.parquet","data/ASR/sna/sna-validation-*.parquet",
  ],
  max_workers=8,
)
print("DL_DONE")