openai_humaneval

References:

• Code
• Huggingface

openai_humaneval

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:openai_humaneval/openai_humaneval')

• Description:

The HumanEval dataset released by OpenAI contains 164 handcrafted programming challenges together with unittests to very the viability of a proposed solution.

• License: MIT
• Version: 1.0.0
• Splits:

Split	Examples
'test'	164

• Features:

{
    "task_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "prompt": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "canonical_solution": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "test": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "entry_point": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}