---
tags:
- pytorch
- commonsense-reasoning
- sentence-completion
datasets:
- hellaswag
---

`RoBERTa` trained on HellaSwag dataset (`MultipleChoiceModel`). HellaSwag has a multiple choice questions format.
It gets around 74.99% accuracy.

[@prajjwal_1](https://twitter.com/prajjwal_1/)