XSTest: A Test Suite for Identifying Exaggerated Safety Behaviours in Large Language Models.
Paul RöttgerHannah KirkBertie VidgenGiuseppe AttanasioFederico BianchiDirk HovyPublished in: NAACL-HLT (2024)
Keyphrases
- language model
- test suite
- language modeling
- test cases
- regression testing
- n gram
- document retrieval
- test suite reduction
- query expansion
- speech recognition
- probabilistic model
- statistical language models
- language models for information retrieval
- language modelling
- information retrieval
- context sensitive
- software testing
- query terms
- retrieval model
- test collection
- document ranking
- mutation testing
- ad hoc information retrieval
- vector space model
- pseudo relevance feedback
- passage retrieval
- document length
- database
- spoken term detection
- feature selection
- number of test cases
- text classification
- cross lingual