AMALIA-VL: New open-source vision-language model for European Portuguese released

Researchers have introduced AMALIA-VL, a new open-source vision and language model specifically designed for European Portuguese. Existing models often overlook or misrepresent this language variant. AMALIA-VL utilizes a high-resolution vision encoder and a language model optimized for European Portuguese, developed through a three-stage training process. The project aims to democratize the development of European Portuguese large vision and language models by releasing model weights, training data, and evaluation benchmarks. AI

IMPACT This release provides a dedicated resource for European Portuguese, potentially improving multimodal AI capabilities for this language group.

RANK_REASON The cluster describes the release of a new open-source model and associated data for a specific language variant, detailed in an arXiv paper. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• AMALIA-VL
• arXiv
• Diogo Glória-Silva
• European Portuguese
• Hugging Face

• model release
• paper

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →