Skip to main content

Fine-grained Hallucination Detection and Editing for Language Models

Β· 2 min read
Zain Hasan

A preview of the paper

A breakdown of the different types of hallucinations from AI2:πŸ„β€‹

  1. Verifiably Factually Wrong ❌
  • Entity: an entity in a statement is incorrect (eg. Christmas falls on Nov. 25th)

  • Relation: semantic relationship in a statement is incorrect (eg. The mouse ate the cat.)

  • Contradictory: statements that entirely contradict relevant evidence from the web (eg. Raptors are yet to win an NBA final.)

  1. Unverifiable Types of Hallucinations ⁉️
  • Invented: statements of concepts that do not exist in world knowledge (eg. MJ created the sideways somersault)

  • Subjective: Statement that lacks universal validity - basically an opinion (eg. The Raptors are the best NBA team)

  • Unverifiable: potentially factual statement but cannot be grounded in world evidence(eg. Jensen sleeps in a leather jacket.)

πŸ”Word vs. Sentence Level:​

Entity and Relation are usually word level, and so can be fixed with small edits if you know where they occur.

Contradictory, Invented, Subjective, and Unverifiable are often sentence level and thus need to be removed completely to fix the issue.

πŸ’»Code

πŸ”— arXiv Link

πŸ“œ Download paper

Ready to start building?​

Check out the Quickstart tutorial, and begin building amazing apps with the free trial of Weaviate Cloud (WCD).

Don't want to miss another blog post?

Sign up for our bi-weekly newsletter to stay updated!


By submitting, I agree to the Terms of Service and Privacy Policy.