SSL certificate verification failure when loading `HFDataset` locally

[ElenaKhaustova]

Description

When loading HFDataset using datasets.load_dataset() (via HFDataset.load()), SSL verification fails in some local environments:

httpx.ConnectError: [SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed: self-signed certificate in certificate chain

This issue does not affect CI after doctest skipping (see #1293), but remains relevant for local developer experience and documentation.

Context

This occurs on macOS with Python 3.11, 3.12, and 3.13. The issue is independent of Kedro and reproduces when calling datasets.load_dataset() directly.

Steps to Reproduce

1. pip install "kedro-datasets[huggingface-hfdataset]"
2. Run

from kedro_datasets.huggingface import HFDataset

dataset_name = "openai_humaneval"
dataset = HFDataset(dataset_name=dataset_name)
ds = dataset.load()

3. Or run without Kedro dataset

from datasets import load_dataset

ds = load_dataset(
    "openai_humaneval",
)

print("Keys in dataset:", ds.keys())
print("Train examples:", len(ds["train"]))

Notes

• Not caused by Kedro implementation
• Not fixed by downgrading datasets or huggingface_hub
• It worked a few times when turning off VPN
• Likely requires explicit CA trust configuration or documentation