{
    "class": "EmuDataset",
    "dataset_path": "/workspace/hygon/K72",
    "num_samples": 224000,
    "index_split": "train",
    "random_seed": 19971104,
    "sequence_length": 32768,
    "split": null,
    "split_matrix": null,
    "tokenizer": {
        "class": "_Emu3TokenizerFS",
        "tokenizer_path": [
            "HFTokenizer"
        ]
    }
}