cnmoro's picture
Update README.md
8844821
metadata
license: mit
language:
  - pt

Model based on:

tgsc/ult5-pt-small

Finetuned on a self-made instruction dataset.

Portuguese only.

Trained on input context length: 1024 tokens

output context length: 512

But generally, T5 models accepts more than the default input context length.