INSAIT-Institute/MamayLM-Gemma-3-27B-IT-v2.0
The INSAIT-Institute/MamayLM-Gemma-3-27B-IT-v2.0 is a 27 billion parameter instruction-tuned vision-language model developed by INSAIT, based on the Gemma 3 architecture. This model is specifically adapted for Ukrainian language and culture, offering improved localization and an updated knowledge cut-off to October 2025. It excels in understanding both text and images within the same context, making it suitable for multimodal applications requiring strong instruction-following capabilities in Ukrainian.
Loading preview...
MamayLM-Gemma-3-27B-IT-v2.0 Overview
This model is a 27 billion parameter, instruction-tuned vision-language model from INSAIT, built upon the Gemma 3 architecture. It represents the second version of the MamayLM series, specifically enhanced for Ukrainian language and cultural understanding.
Key Capabilities and Improvements
- Vision-Language Understanding: Processes and understands both textual and image inputs within the same context, enabling multimodal interactions.
- Enhanced Instruction-Following: Trained on a diverse range of tasks, supporting multi-turn conversations, complex instructions, and system prompts more effectively.
- Improved Localization: Demonstrates better alignment with Ukrainian linguistic nuances and cultural contexts compared to its predecessor.
- Updated Knowledge Base: Incorporates pretraining data up to May 2025 and instruction fine-tuning data up to October 2025, providing more current information.
Use Cases and Considerations
This model is particularly well-suited for applications requiring robust Ukrainian language processing combined with multimodal capabilities. Its instruction-following improvements make it effective for chatbots, content generation, and complex query resolution in Ukrainian. Developers should note the availability of an FP8 dynamic quantized variant for reduced memory footprint while preserving vision capabilities, as direct FP8 quantization via vLLM's on-the-fly flag is not recommended for this model due to potential vision degradation.