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
updated a Space 3 days ago
Papuguinho/README liked a Space 4 days ago
perplexity-ai/q2d-web-leaderboard updated a dataset 17 days ago
tardellirs/enem-audiodescricaoOrganizations
Contamination-clean ~110M Brazilian-Portuguese embeddings (BERTimbau). #1 in the ~100M class on MTEB(por).
-
tardellirs/brazembed-pt-br
Sentence Similarity • 0.1B • Updated • 111 • 1 -
tardellirs/brazembed-pt-br-clfclu
Sentence Similarity • 0.1B • Updated • 101 -
tardellirs/brazembed-pt-br-pairclf
Sentence Similarity • 0.1B • Updated • 78 -
tardellirs/brazembed-pt-br-sts
Sentence Similarity • 0.1B • Updated • 79
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 • 2.47k • 1 -
tardellirs/embeddinggemma-pt-br-128k
Sentence Similarity • 0.2B • Updated • 76 • 1 -
tardellirs/embeddinggemma-pt-br-48k
Sentence Similarity • 0.1B • Updated • 75 • 1 -
tardellirs/embeddinggemma-pt-br-32k
Sentence Similarity • 0.1B • Updated • 79 • 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 • 111 • 1 -
tardellirs/brazembed-pt-br-clfclu
Sentence Similarity • 0.1B • Updated • 101 -
tardellirs/brazembed-pt-br-pairclf
Sentence Similarity • 0.1B • Updated • 78 -
tardellirs/brazembed-pt-br-sts
Sentence Similarity • 0.1B • Updated • 79
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 • 2.47k • 1 -
tardellirs/embeddinggemma-pt-br-128k
Sentence Similarity • 0.2B • Updated • 76 • 1 -
tardellirs/embeddinggemma-pt-br-48k
Sentence Similarity • 0.1B • Updated • 75 • 1 -
tardellirs/embeddinggemma-pt-br-32k
Sentence Similarity • 0.1B • Updated • 79 • 2