Dataset · Rowan
HellaSwag Commonsense Benchmark
Rowan/hellaswag
Multiple-choice sentence completion tasks to evaluate commonsense reasoning in language models.
What's inside
English commonsense NLI dataset where models choose a plausible continuation for a given situation. Published at ACL 2019; metadata indicates between 10k and 100k examples. File sizes are approximately 71.49 MB with a generated set of 65.32 MB.
What you can do with it
- Evaluating LLM commonsense reasoning
- Benchmark for NLI models
- Assessing text continuation selection
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 282.2K
- Likes
- 217
Models trained on it
- mradermacher/Atomight-V2.2-0.5B-UltraThink-GGUF 559
- mradermacher/Atomight-V2.2-UltraThink-0.5B-GGUF 351
- decision-tune/decisiontune-1.0Zero shot classification 309
- usermma/Atomight-V2.2-0.5B-UltraThink-GGUF-mradermacher 198
- Tractorman/accel-lm-5M-GGUF 174
- Danny-Dasilva/embeddinggemma-2-multimodal-typed-decisionZero shot classification 152
- whittleOrg/Whittle-Reflex-25B-A3BZero shot classification 134
- dawncr0w/CrowMuse-Glimmer-30B-HereticImage text → text 106
How to load it
from datasets import load_dataset
ds = load_dataset("Rowan/hellaswag")