Dataset · VISAI-AI
Thai legal question and answer benchmark
VISAI-AI/nitibench-plus
Questions and answers based on real Thai court rulings with reasoning and legal references.
What's inside
A benchmark ranging from 1,000 to 10,000 records in Thai, gathered from decisions of the Supreme Court, Supreme Administrative Court, and Constitutional Court of Thailand. Each question is accompanied by the court verdict, justification, and law articles as amended on the ruling date.
What you can do with it
- Evaluating legal accuracy of LLMs and question-answering systems in Thai
- Testing RAG pipelines on the ability to reference law versions active at the time of the event
- Tasks
- Question answering
- Languages
- Thai
- Size
- 1K–10K rows
- Downloads this month
- 3.8K
- Likes
- 2
- Access
- on request on Hugging Face
How to load it
from datasets import load_dataset
ds = load_dataset("VISAI-AI/nitibench-plus")