truthfulqa/truthful_qa
Multiple ChoiceText GenerationQuestion AnsweringENBenchmark
Truthfulqa/truthful_qa is a multiple choice-focused benchmark dataset in EN distributed in Parquet format.
📊 This dataset is used as an LLM benchmark. See model leaderboards →
About truthfulqa/truthful_qa
Dataset Card for truthful_qa
Dataset Summary
TruthfulQA is a benchmark to measure whether a language model is truthful in generating answers to questions. The benchmark comprises 817 questions that span 38 categories, including healt...
Details
- Task
- Multiple Choice, Text Generation, Question Answering
- Language
- EN
- Format
- Parquet
- Rows / instances
- N/A
- Creator
- truthfulqa
- Year
- 2022