Modelpublic

tardellirs/brazembed-pt-br-clfclu

sourceHugging Facemitupdated 3mo agoView on Hugging Face

0likes101downloads

Model Card

brazembed-pt-br-clfclu — BrazEmbed-PT-BR component (Classification & Clustering)

A component of [BrazEmbed-PT-BR](https://huggingface.co/tardellirs/brazembed-pt-br), the contamination-clean ~110M Brazilian-Portuguese embedding system (task-routed, MTEB(por) mean_16 = 0.6567, #1 in the ~100M class). This standalone SentenceTransformer (Brazilian BERTimbau + the Classification & Clustering clean weight-soup) serves the Classification & Clustering tasks.

python

from sentence_transformers import SentenceTransformer
m = SentenceTransformer("tardellirs/brazembed-pt-br-clfclu")   # mean-pooling, L2-normalized, no instruction prefix

Use it directly for Classification & Clustering, or via the router (https://github.com/tardellirs/brazembed-pt-br → route.py). For one general model, use tardellirs/brazembed-pt-br. License MIT. Benchmark: MTEB(por) (leaderboard).