How can I see the size of a HuggingFace dataset before downloading it?

I want to download a HuggingFace dataset, e.g. uonlp/CulturaX:

from datasets import load_dataset
ds = load_dataset("uonlp/CulturaX", "en")

How can I see the size of a HuggingFace dataset before downloading it?

One can use HfApi.api.dataset_info from huggingface_hub:

from huggingface_hub import HfApi

def print_dataset_file_sizes(repo_id):
    api = HfApi()
    dataset_info = api.dataset_info(repo_id=repo_id, files_metadata=True)

    total_size_bytes = 0  
    print(f"File sizes for dataset '{repo_id}':\n")  
    for sibling in dataset_info.siblings:  
        filename = sibling.rfilename  
        size_in_bytes = sibling.size or 0  
        total_size_bytes += size_in_bytes  
        size_mb = size_in_bytes / (1024 * 1024)  
        print(f"  {filename}: {size_mb:.2f} MiB")  

    total_size_mb = total_size_bytes / (1024 * 1024)  
    print(f"\nTotal size: {total_size_mb:.2f} MiB")  

print_dataset_file_sizes('uonlp/CulturaX')

Outputs: Total size: 16658008.77 MiB, i.e. ~15.9 TiB.

Note that the same code can be used to see the size of a HuggingFace model before downloading it, just replace HfApi.api.dataset_info with HfApi.api.model_info, e.g.:

from huggingface_hub import HfApi

def print_dataset_file_sizes(repo_id):
    api = HfApi()
    dataset_info = api.model_info(repo_id=repo_id, files_metadata=True)

    total_size_bytes = 0  
    print(f"File sizes for dataset '{repo_id}':\n")  
    for sibling in dataset_info.siblings:  
        filename = sibling.rfilename  
        size_in_bytes = sibling.size or 0  
        total_size_bytes += size_in_bytes  
        size_mb = size_in_bytes / (1024 * 1024)  
        print(f"  {filename}: {size_mb:.2f} MiB")  

    total_size_mb = total_size_bytes / (1024 * 1024)  
    print(f"\nTotal size: {total_size_mb:.2f} MiB")  

print_dataset_file_sizes('deepseek-ai/DeepSeek-V3')

outputs: Total size: 656703.88 MiB (i.e., ~641.3 GiB).