analiza aiqbench
This study addresses the critical gap in evaluating text embedding models specifically for Portuguese, demonstrating that performance metrics derived from English or multilingual averages are unreliable proxies. The research reveals strong task-dependency and highlights the necessity of language-specific fine-tuning to achieve optimal results across semantic similarity, classification, and retrieval tasks. By releasing a dedicated benchmark and fine-tuned baselines using Matryoshka Representation Learning, this work provides essential infrastructure for developers building robust NLP systems in Portuguese-speaking regions.
Beyond Multilingual Averages: MTEB-PT, a Benchmark for Portuguese Sentence Encoders
arXiv:2607.04071v2 Announce Type: replace-cross Abstract: Portuguese remains underrepresented in text embedding evaluation, despite being one of the most widely spoken languages in the world. As a result, embedding models are often selected based on English or multilingual metrics, while their effectiveness in Portuguese remains unclear. We present MTEB-PT, a Portuguese benchmark constructed from a subset of MMTEB, comprising 14 existing datasets across Semantic Textual Similarity (STS), classification, retrieval, and reranking. We use this benchmark to evaluate 17 open- and closed-source embedding models under a unified protocol. Our results show that Portuguese performance is strongly task-dependent: multilingual rankings do not reliably predict Portuguese-specific performance across task families, no single model dominates all settings, and models with stronger long-context capacity are particularly advantageous on longer-input tasks such as retrieval and reranking. The benchmark also shows that language-specific fine-tuning still improves model performance in Portuguese, especially on task types that match the adaptation data most closely. To examine this effect, we fine-tune three representative backbone models with Portuguese contrastive supervision and Matryoshka Representation Learning (MRL). These benchmark-informed baselines yield their strongest gains on STS, consistent with the predominantly symmetric supervision used during training, while also improving retrieval and remaining competitive under dimensional truncation. We release the MTEB-PT benchmark, the fine-tuned models, and the training and evaluation code.
Abstract:Portuguese remains underrepresented in text embedding evaluation, despite being one of the most widely spoken languages in the world. As a result, embedding models are often selected based on English or multilingual metrics, while their effectiveness in Portuguese remains unclear. We present MTEB-PT, a Portuguese benchmark constructed from a subset of MMTEB, comprising 14 existing datasets across Semantic Textual Similarity (STS), classification, retrieval, and reranking. We use this benchmark to evaluate 17 open- and closed-source embedding models under a unified protocol. Our results show that Portuguese performance is strongly task-dependent: multilingual rankings do not reliably predict Portuguese-specific performance across task families, no single model dominates all settings, and models with stronger long-context capacity are particularly advantageous on longer-input tasks such as retrieval and reranking. The benchmark also shows that language-specific fine-tuning still improves model performance in Portuguese, especially on task types that match the adaptation data most closely. To examine this effect, we fine-tune three representative backbone models with Portuguese contrastive supervision and Matryoshka Representation Learning (MRL). These benchmark-informed baselines yield their strongest gains on STS, consistent with the predominantly symmetric supervision used during training, while also improving retrieval and remaining competitive under dimensional truncation. We release the MTEB-PT benchmark, the fine-tuned models, and the training and evaluation code.
Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
arXiv:2607.04071 [cs.CL]
(or arXiv:2607.04071v2 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2607.04071 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 308 cuvinte
Subiecte conexe
Granite Embedding Multilingual R2: Open Apache 2.0 Multilingual Embeddings with 32K Context — Best Sub-100M Retrieval Quality
Hugging Face · 14 mai 2026
Multi-Vector (Late Interaction) Embedding Models with Sentence Transformers
Hugging Face · 18 aug. 2026
Introducing the Ettin Reranker Family
Hugging Face · 19 mai 2026
Training and Finetuning Multi-Vector Embedding Models with Sentence Transformers
Hugging Face · 26 aug. 2026