Tiny-Aya-Water Blind Spot Evaluation Dataset
Published:
This dataset contains 10 manually constructed blind spot cases designed to evaluate reasoning, instruction following, and factual robustness of the model:
Christophe D. Hounwanou
Ph.D Student in Computer Science at Mila, IID & Université Laval
- Montreal, Quebec, Canada
- Université Laval, IID & Mila
- Google Scholar
- ORCID
- ResearchGate
- Github
- X (formerly Twitter)