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 dataset about 7 hours ago
tardellirs/enem-audiodescricao published a dataset about 8 hours ago
tardellirs/enem-audiodescricao updated a dataset about 8 hours ago
tardellirs/enem-audiodescricaoOrganizations
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 • 1.6k • 1 -
tardellirs/embeddinggemma-pt-br-128k
Sentence Similarity • 0.2B • Updated • 90 • 1 -
tardellirs/embeddinggemma-pt-br-48k
Sentence Similarity • 0.1B • Updated • 93 • 1 -
tardellirs/embeddinggemma-pt-br-32k
Sentence Similarity • 0.1B • Updated • 92 • 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).
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 • 1.6k • 1 -
tardellirs/embeddinggemma-pt-br-128k
Sentence Similarity • 0.2B • Updated • 90 • 1 -
tardellirs/embeddinggemma-pt-br-48k
Sentence Similarity • 0.1B • Updated • 93 • 1 -
tardellirs/embeddinggemma-pt-br-32k
Sentence Similarity • 0.1B • Updated • 92 • 2