Instructions to use google/umt5-base with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use google/umt5-base with Transformers:
# Load model directly from transformers import AutoModelForSeq2SeqLM model = AutoModelForSeq2SeqLM.from_pretrained("google/umt5-base", dtype="auto") - Notebooks
- Google Colab
- Kaggle
Add a Better Transformer api ?
#3
by 1TuanPham - opened
I think the model deserve this, since without it, the model is really hard to train with limited hardware.
Hi @1TuanPham
We are going to migrate Better Transformer API to transformers core soon.
Please have a look at :https://github.com/huggingface/transformers/pull/26572
Meanwhile I suggest you to train the model using adapters for more memory efficient training: https://huggingface.co/docs/transformers/peft