Dataset.from_dict() can't handle large dict

[CSU-OSS]

Describe the bug

I have 26,000,000 3-tuples. When I use Dataset.from_dict() to load, neither. py nor Jupiter notebook can run successfully. This is my code:

    # len(example_data) is 26,000,000, 'diff' is a text
    diff1_list = [example_data[i].texts[0] for i in range(len(example_data))]
    diff2_list = [example_data[i].texts[1] for i in range(len(example_data))]
    label_list = [example_data[i].label for i in range(len(example_data))]

    embedding_dataset = Dataset.from_dict({
        "diff1": diff1_list,
        "diff2": diff2_list,
        "label": label_list
    })

Steps to reproduce the bug

1. Initialize a large 3-tuple, e.g. 26,000,000
2. Use Dataset.from_dict() to load

Expected behavior

Dataset.from_dict() run successfully

Environment info

sentence-transformers 3.3.1