metadata
language:
- da
license: apache-2.0
daT5-large
A smaller version of Google's mt5-large model, where the original model is reduced to only include Danish embeddings.
How to use
from transformers import AutoTokenizer, AutoModel
tokenizer = AutoTokenizer.from_pretrained("emillykkejensen/daT5-large")
model = AutoModel.from_pretrained("emillykkejensen/daT5-large")
Further reading
Gist showing (in Danish) how the embeddings are extracted (for mt5-base)
Article explaining how to do it by David Dale