Skip to main content

Hallucination Guard

The Hallucination Guard is an output guard that analyzes the responses generated by your language model to detect any fabricated or inaccurate information, ensuring all outputs are factually correct and reliable.

info

HallucinationGuard is only available as an output guard.

Example​

from deepeval.guardrails import HallucinationGuard

model_output = "The capital of Australia is Sydney."

hallucination_guard = HallucinationGuard()
guard_result = hallucination_guard.guard(response=model_output)

There are no required arguments when initializing the HallucinationGuard object. The guard function accepts a single parameter response, which is the output of your LLM application.

Interpreting Guard Result​

print(guard_result.score)
print(guard_result.score_breakdown)

guard_result.score is an integer that is 1 if the guard has been breached. The score_breakdown for HallucinationGuard is a dictionary containing:

  • score: A binary value (1 or 0), where 1 indicates that hallucinated content was detected.
  • reason: A brief explanation of why the score was assigned.
{
  "score": 1,
  "reason": "The statement 'The capital of Australia is Sydney' is incorrect; the capital is Canberra."
}