HallucinationDetectionScorer
Detects hallucinations and factual inconsistencies in generated answers by comparing against context. Identifies specific hallucination types and provides examples.
Overview
Detects hallucinations and factual inconsistencies in generated answers by comparing against context. Identifies specific hallucination types and provides examples.
Use Cases
- RAG-based question answering systems
- Safety and compliance monitoring
- Hallucination detection in generated content
How It Works
This scorer uses LLM-as-Judge technology to evaluate responses. It prompts a large language model with specific evaluation criteria and the content to assess, then analyzes the LLM's judgment to produce a score and detailed reasoning.
Input Schema
| Parameter | Type | Required | Description |
|---|---|---|---|
| output_text | str | Yes | Answer to check for hallucinations |
| input_text | str | No | Original question |
| context | dict | str | list | Yes | Context to verify against |
Output Schema
| Field | Type | Description |
|---|---|---|
| score | float | Non-hallucination score (0-10, 10=no hallucination) |
| passed | bool | True if no hallucinations |
| reasoning | str | Hallucination analysis |
| metadata.hallucination_types | list | Types of hallucinations found |
| metadata.specific_examples | list | Specific hallucination examples |
Score Interpretation
Default threshold: 7/10
Related Scorers
Frequently Asked Questions
When should I use this scorer?
Use HallucinationDetectionScorer when you need to evaluate hallucination and rag aspects of your AI outputs. It's particularly useful for rag-based question answering systems.
Why doesn't this scorer need expected output?
This scorer evaluates quality aspects that don't require comparison against a reference answer. It uses the system prompt and context as the implicit ground truth.
Can I customize the threshold?
Yes, the default threshold of 7 can be customized when configuring the scorer.
Quick Info
Ready to try HallucinationDetectionScorer?
Start evaluating your AI agents with Noveum.ai's comprehensive scorer library.
Explore More Scorers
Discover 68+ LLM-as-Judge scorers for comprehensive AI evaluation
