| Check Type | Description |
|---|---|
| Toxicity Scoring | Score output for hate speech, abuse, profanity |
| Hallucination Detection | Compare output claims against source documents |
| PII in Output | Ensure model didn't leak sensitive training data |
| Factual Consistency | Cross-check key claims against verified data |
| Format Validation | Ensure output matches expected schema (JSON, XML, etc.) |
| Relevance Check | Ensure response addresses the original question |
Created
March 16, 2026 19:24
-
-
Save shubhamkakkar/221c5c01cf8dd70b60dc077cfee75f82 to your computer and use it in GitHub Desktop.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment