The MTEB(por) benchmark for Brazilian Portuguese (leaderboard + task datasets) and the Colibri embedding model that tops it among small models.
Tardelli Stekel
tardellirs
AI & ML interests
None yet
Recent Activity
liked a Space 25 minutes ago
modelpulse/model-pulse posted an update about 1 hour ago
I built Model Pulse: daily download history for every model on the Hub 📈
A model page tells you its downloads for the last 30 days, but not how it got there. Model Pulse rebuilds the day-by-day history for 1.6M models, back to July 2024, from the daily snapshots in @cfahlgren1's hub-stats dataset.
A few things the data shows:
- The Hub now serves ~100M model downloads a day, up from ~62M a year ago
- Vision-language models grew about 7x in that year, to 10.7M downloads a day
- 37% of Qwen3-8B's monthly downloads go to its 6,283 quantizations, fine-tunes and merges, not to the original
What you can do with it:
- Open any model, e.g. https://huggingface.co/spaces/modelpulse/model-pulse?model=Qwen/Qwen3-8B
- Compare up to five models on one chart
- Browse weekly rankings: fastest growing, new this month, biggest families, top organizations
- Add a live badge to your model card (monthly downloads, sparkline and weekly trend)
The data is open too: https://huggingface.co/datasets/modelpulse/model-pulse-data
I'd love feedback, especially from model authors: what would you want to see about your own models? updated a Space about 3 hours ago
modelpulse/READMEOrganizations
Contamination-clean ~110M Brazilian-Portuguese embeddings (BERTimbau). #1 in the ~100M class on MTEB(por).
-
tardellirs/brazembed-pt-br
Sentence Similarity • 0.1B • Updated • 682 • 2 -
tardellirs/brazembed-pt-br-clfclu
Sentence Similarity • 0.1B • Updated • 97 -
tardellirs/brazembed-pt-br-pairclf
Sentence Similarity • 0.1B • Updated • 70 -
tardellirs/brazembed-pt-br-sts
Sentence Similarity • 0.1B • Updated • 106
Compact pt-br text-embedding model (~157M) that beats far larger models on MTEB(por): vocabulary trim -> Qwen3-4B distillation -> model soup.
Portuguese EmbeddingGemma-300M, vocab-trimmed 16k–128k. Training-free; 64k ≈ full quality at half the params.
-
tardellirs/embeddinggemma-pt-br
Sentence Similarity • 0.2B • Updated • 2.48k • 1 -
tardellirs/embeddinggemma-pt-br-128k
Sentence Similarity • 0.2B • Updated • 94 • 1 -
tardellirs/embeddinggemma-pt-br-48k
Sentence Similarity • 0.1B • Updated • 85 • 1 -
tardellirs/embeddinggemma-pt-br-32k
Sentence Similarity • 0.1B • Updated • 77 • 2
The MTEB(por) benchmark for Brazilian Portuguese (leaderboard + task datasets) and the Colibri embedding model that tops it among small models.
Compact pt-br text-embedding model (~157M) that beats far larger models on MTEB(por): vocabulary trim -> Qwen3-4B distillation -> model soup.
Contamination-clean ~110M Brazilian-Portuguese embeddings (BERTimbau). #1 in the ~100M class on MTEB(por).
-
tardellirs/brazembed-pt-br
Sentence Similarity • 0.1B • Updated • 682 • 2 -
tardellirs/brazembed-pt-br-clfclu
Sentence Similarity • 0.1B • Updated • 97 -
tardellirs/brazembed-pt-br-pairclf
Sentence Similarity • 0.1B • Updated • 70 -
tardellirs/brazembed-pt-br-sts
Sentence Similarity • 0.1B • Updated • 106
Portuguese EmbeddingGemma-300M, vocab-trimmed 16k–128k. Training-free; 64k ≈ full quality at half the params.
-
tardellirs/embeddinggemma-pt-br
Sentence Similarity • 0.2B • Updated • 2.48k • 1 -
tardellirs/embeddinggemma-pt-br-128k
Sentence Similarity • 0.2B • Updated • 94 • 1 -
tardellirs/embeddinggemma-pt-br-48k
Sentence Similarity • 0.1B • Updated • 85 • 1 -
tardellirs/embeddinggemma-pt-br-32k
Sentence Similarity • 0.1B • Updated • 77 • 2