Handling Validation Errors
You'll learn: How to gracefully catch Pydantic validation errors and use them to improve your LLM outputs through retries or clear user feedback.
When you parse LLM-generated JSON into Pydantic models, things won't always go smoothly. The model might return invalid data types, missing required fields, or values that violate your custom constraints. Pydantic will raise a ValidationError when this happens—and that's actually a good thing.
Why Validation Errors Are Opportunities
Think of validation errors like a teacher marking a test: they tell you exactly what went wrong. Instead of silently accepting bad data, you can:
- Retry with improved prompts – Include the error details in a follow-up request, asking the LLM to fix its mistakes
- Provide helpful feedback – Show users meaningful error messages instead of cryptic crashes
Catching and Using Validation Errors
When you call .model_validate() or .model_validate_json(), wrap it in a try-except block:
from pydantic import ValidationError
try:
result = MyModel.model_validate_json(llm_response)
except ValidationError as e:
# e.errors() gives structured error details
print(f"Validation failed: {e}")
# Option 1: Retry with error context
retry_prompt = f"Previous attempt failed: {e}. Please correct and retry."
# Option 2: Inform the user
return {"error": "Invalid data format", "details": e.errors()}
The ValidationError object contains .errors() with field-specific details—perfect for building a corrective prompt or explaining the problem to users.
Key Takeaway: Catching Pydantic validation errors lets you turn LLM mistakes into retry opportunities or user-friendly feedback, making your applications more robust and resilient.