The Benchmark of Linguistic Minimal Pairs (BLiMP)
Language ModelingEnglish
The Benchmark of Linguistic Minimal Pairs (BLiMP) is a language modeling dataset in English from Warstadt et al. with 67 sub-datasets each with 1,000 minimal pairs records in JSON format.
About The Benchmark of Linguistic Minimal Pairs (BLiMP)
BLiMP is a challenge set for evaluating what language models (LMs) know about major grammatical phenomena in English.
Details
- Task
- Language Modeling
- Language
- English
- Format
- JSON
- Rows / instances
- 67 sub-datasets each with 1,000 minimal pairs
- Creator
- Warstadt et al.
- Year
- 2019