A custom model2vec model, trained using a modified version of the
tokenlearn
library.
The output dimension is 256, and the vocabulary size is 10.000.
The training process used a mix of English (10%) and Portuguese (90%) texts.
from model2vec import StaticModel
# Load a pretrained Sentence Transformer model
model = StaticModel.from_pretrained("cnmoro/custom-model2vec-tokenlearn-small")
# Compute text embeddings
embeddings = model.encode(["Example sentence"])