LUMAMODEL/luma
LUMA is a compact 0.5 billion parameter instruction-tuned causal language model developed by LUMAMODEL, based on Qwen2.5-0.5B-Instruct. It is specifically fine-tuned for natural, human-like Russian dialogue and conversational assistance, capable of understanding context and generating code. With a 32768 token context length, it is optimized for interactive chat experiences in Russian and can run efficiently even on CPU.
Loading preview...
LUMA: Your Friendly Russian Assistant
LUMA is a compact, instruction-tuned language model developed by LUMAMODEL, built upon the Qwen2.5-0.5B-Instruct architecture. With 0.5 billion parameters and a 32768 token context length, it is specifically designed for natural and engaging Russian dialogue. The model was fine-tuned using LoRA on a corpus of Russian dialogues, with the adapter weights merged into the full model for immediate use.
Key Capabilities
- Natural Russian Dialogue: Excels at human-like conversations, understanding context, and maintaining coherence.
- Code Generation: Capable of generating code and HTML when prompted, with specific system prompt recommendations for full, valid output.
- Efficient Operation: Its small size (494 million parameters) allows it to run effectively even on CPU.
- Free and Open: Distributed under the Apache-2.0 license.
Good For
- Conversational AI in Russian: Ideal for chatbots and virtual assistants requiring fluent and context-aware Russian interaction.
- Educational Tools: Can assist with coding tasks or provide explanations in Russian.
- Resource-Constrained Environments: Suitable for deployment where computational resources are limited, such as local CPU inference.