ahmedabdelwahed
/

Mojiz-DPO-1e-5-4000-steps-beta-1e-1

Create README.md

b21970c 8 months ago

159 Bytes

metadata

license: mit
language:
  - ar
library_name: transformers
tags:
  - summarization

This model was aligned using DPO with a 1e-5 learning rate for 4000 steps