This
Model2Vec model was created by using
Tokenlearn, with
nomic-embed-text-v2-moe as a base, trained on around 20M passages (english and portuguese).
I have yet to run any benchmarks on it, but it easily outperforms
potion-multilingual-128M on my custom-portuguese-testing-workload-thing.
The output dimension is 768.
1from model2vec import StaticModel
2
3# Load a pretrained Model2Vec model
4model = StaticModel.from_pretrained("cnmoro/static-nomic-eng-ptbr-large")
5
6# Compute text embeddings
7embeddings = model.encode(["Example sentence"])
