---
tags:
- summarization
- bart
language:
- ro
inference: false
---

This is a pretrained-from-scratch **BART base** model (**140M** parameters).

Training was performed on a clean **50GB Romanian** text corpus for 3M steps with these [scripts](https://github.com/cosmoquester/transformers-bart-pretrain). The model was trained with a maximum sequence length of **1024**. 

**!! IMPORTANT !!** This model was pretrained on the text corruption task, meaning this model is **not usable** in any downstream task **without finetuning** first!