Dataset · aps
SuperGLUE language understanding benchmark
aps/super_glue
Challenging NLU tasks including textual entailment, QA, and coreference resolution for language model evaluation.
What's inside
SuperGLUE is a GLUE-style benchmark featuring harder language understanding tasks such as textual entailment, question answering, and coreference resolution. English data, ranging from 100k to 1M examples in tabular and text formats. Includes a public leaderboard.
What you can do with it
- Evaluate LLMs and encoders on NLU tasks
- Fine-tune BERT-style models for classification and NER
- Compare models via a unified leaderboard
- Tasks
- Text classification, Token classification, Question answering
- Languages
- English
- Size
- 100K–1M rows
- Downloads this month
- 202.6K
- Likes
- 213
Models trained on it
- tasksource/deberta-small-long-nliZero shot classification 9.3K
- tasksource/deberta-base-long-nliZero shot classification 1.4K
- mradermacher/elythros-decider-1.7b-GGUFZero shot classification 887
- samratduttaofficial/WaterSheepZero shot classification 87
- Elythros/elythros-decider-4b-expText classification 85
- cglez/bert-base-uncased-ft-rteText classification 84
- Elythros/elythros-decider-1.7bText classification 53
- zeroMN/SHMTAudio text → text 40
How to load it
from datasets import load_dataset
ds = load_dataset("aps/super_glue")