Hugging Face Datasets

RIME now offers a data provider that natively integrates with Hugging Face’s load_dataset API. To use this functionality, specify a HuggingFace data configuration.

Data Info Configuration Example

{
    "connection_info": {
        "hugging_face": {
            "dataset_uri": "sst2",
            "split_name": split_name,
            "loading_params_json": json.dumps({}),
        }
    },
    "data_params": {
        "label_col": "label",
        "text_features": ["sentence"],
        "sample": True,
        "nrows": 10
        "class_names": ["Nimitz", "Arkleigh", "Zumwalt"],
    },
}