HiTZ/Latxa-Qwen3.5-2B
HiTZ/Latxa-Qwen3.5-2B is a 2.3 billion parameter multimodal and multilingual instruct model developed by HiTZ Research Center & IXA Research group, built upon Qwen3.5-2B. This vision-language LLM is specifically adapted for enhanced performance in Basque, Galician, and Catalan, while also supporting Spanish and English. It excels at understanding and generating text, processing images, and following interactive instructions, making it suitable for applications requiring robust low-resource language capabilities.
Loading preview...
Model Overview
HiTZ/Latxa-Qwen3.5-2B is a 2.3 billion parameter multimodal and multilingual instruct model developed by the HiTZ Research Center & IXA Research group. It is built on the Qwen3.5-2B architecture, a powerful vision-language LLM capable of understanding and generating text, and processing images. The model has been specifically adapted for improved performance in Basque, Galician, and Catalan, in addition to supporting Spanish and English.
Key Capabilities & Features
- Multimodal: Processes both text and image inputs, making it suitable for vision-language tasks.
- Multilingual Adaptation: Significantly enhanced for Basque, Galician, and Catalan, addressing low-resource language challenges.
- Instruction Following: Designed to follow instructions and function as a chat assistant.
- Performance: Demonstrates substantial improvements over the base Qwen3.5-2B model across various Basque-specific benchmarks, with gains up to +40.11% on BL2MP and +26.68% on Arc Easy for the 'multi' variant.
Use Cases
This model is primarily intended for applications involving Basque, Galician, and Catalan data, particularly for instruction-following and chat assistant roles. Its multimodal capabilities also make it suitable for tasks requiring both visual and textual understanding in these languages. Users should note that performance is not guaranteed for other languages, and it is not intended for malicious activities.