For the complete documentation index, see llms.txt. This page is also available as Markdown.

Valid CSV

Ensure AI outputs produce correct CSV structure. Detect errors before data is processed or used

Objective Validates if the text is in a proper CSV format, ensuring data integrity.

Interpretation A higher score indicates the text is not valid CSV. A lower (or zero) score indicates the response is valid CSV.

Code Execution

metrics = [
    {
        "name": "Valid CSV",
        "config": {
            "model": "gpt-4o-mini",
            "provider": "openai"
        },
        "column_name": "your-column-identifier",
        "schema_mapping": schema_mapping
    }
]

Example

  • Prompt: “Provide a CSV of three fruit names and their colors.”

  • Context: “The CSV must have two columns: Fruit,Color.”

  • Response:

    mathematicaCopyApple, Red
    Banana Yellow
    Cherry, Red
  • Metric Output: {"score": 1, "reason": "Invalid CSV format (missing comma in second line)."}

Last updated

Was this helpful?