INSAIT-Institute/MamayLM-Gemma-3-12B-IT-v2.0
MamayLM-Gemma-3-12B-IT-v2.0 is a 12 billion parameter instruction-tuned large language model developed by INSAIT-Institute, based on the Gemma 3 architecture. This model is specifically adapted for Ukrainian language and culture, offering enhanced localization and an updated knowledge cut-off to October 2025. It features vision-language understanding, allowing it to process both text and images within the same context, and demonstrates improved instruction-following capabilities for complex and multi-turn conversations.
Loading preview...
MamayLM-Gemma-3-12B-IT-v2.0 Overview
MamayLM-Gemma-3-12B-IT-v2.0 is a 12 billion parameter instruction-tuned model from INSAIT-Institute, building upon the Gemma 3 architecture. It represents the second iteration of the MamayLM series, specifically designed with a strong focus on Ukrainian language and cultural understanding. This version significantly improves upon its predecessor by integrating multimodal capabilities, allowing it to interpret both text and images.
Key Capabilities
- Vision-Language Understanding: The model can process and understand information from both text and image inputs simultaneously within the same context.
- Enhanced Instruction-Following: It is trained to handle a wider array of tasks, including multi-turn conversations, complex instructions, and system prompts more effectively.
- Improved Localization: Demonstrates better alignment and understanding of the Ukrainian language and cultural nuances.
- Updated Knowledge Base: Features an extended knowledge cut-off, with pretraining data up to May 2025 and instruction fine-tuning data up to October 2025.
Good For
- Applications requiring strong performance in the Ukrainian language.
- Use cases that benefit from multimodal input, combining text and image understanding.
- Complex conversational AI and instruction-based tasks where nuanced understanding is critical.