leap-stc / ClimSim

An open large-scale dataset for training high-resolution physics emulators in hybrid multi-scale climate simulators.
https://leap-stc.github.io/ClimSim/
Apache License 2.0
140 stars 40 forks source link

Get quickstart to download straight from HuggingFace #72

Open jerrylin96 opened 10 months ago

jerrylin96 commented 10 months ago

Quickstart assumes data is locally downloaded, which may not be ideal for some users.

cisaacstern commented 10 months ago

Good call.

I've never used it but have heard good things about https://pypi.org/project/pooch/ for this use case.

jackmiller2003 commented 8 months ago

When I was downloading the data I used huggingface_hub although this may add too many additional dependencies. I'm not sure if this can be done with pooch but hub allows for really easy download of subsets. For example, here is a script that I used to get just one day's worth of data:

import datasets
import logging
from pathlib import Path
from huggingface_hub import snapshot_download

logging.basicConfig(level=logging.INFO)

custom_cache_dir = ""

datasets.config.DOWNLOADED_DATASETS_PATH = Path(custom_cache_dir)

# Download the dataset from the Hugging Face Hub
download_path = snapshot_download(
    repo_id="LEAP/ClimSim_low-res",
    allow_patterns="train/0001-02/*0001-02-01-*.nc",
    cache_dir=custom_cache_dir,
    repo_type="dataset",
)

print(f"Downloaded the dataset to {download_path}")