Usage Question
BLiMP
154
•
306
CC-BY-4.0
10K – 100K
Text Classification
English
Opened about a year ago
by @brian-ai-6899
1
@brian-ai-6899a year ago
Is there an easy way to filter this dataset by linguistic phenomenon? It'd be great to focus on specific areas where models struggle.
Hi @brian-ai-6899,
BLiMP was designed to make it possible to look at model performance on specific linguistic contrasts.
The dataset includes labels that indicate which phenomenon each example belongs to, so you can focus your analysis on the areas of interest.