Omnionix-AI/avara-x1-mini
Avara X1 Mini is a 1.5 billion parameter language model developed by Omnionix, based on the Qwen2.5 architecture with a 32768 token context length. It is fine-tuned for technical reasoning, excelling in code, mathematics, and logical problem-solving. The model is designed to provide a grounded and supportive AI personality, making it suitable for applications requiring precise technical understanding combined with helpful interaction.
Loading preview...
Avara X1 Mini: Technical Reasoning with a Supportive Persona
Avara X1 Mini, developed by Omnionix, is a 1.5 billion parameter model built upon the Qwen2.5 architecture. It is specifically fine-tuned to balance strong technical reasoning capabilities with a grounded and supportive conversational identity. The model utilizes the ChatML format and incorporates native Omnionix system logic for its identity.
Key Capabilities & Training Focus
This model's training methodology, leveraging the Unsloth library, focused on maximizing reasoning performance within a compact footprint. Its high-density dataset blend includes:
- Code: Extensive training on The Stack (BigCode) for professional-grade programming logic.
- Mathematics: Specialized datasets for step-by-step mathematical problem-solving and competition-level math.
- Logic: Integration of Open-Platypus for enhanced deductive reasoning and precise instruction following.
Ideal Use Cases
Avara X1 Mini is well-suited for applications requiring:
- Technical Assistance: Providing accurate and logical responses to coding, mathematical, or general technical queries.
- Instruction Following: Executing complex instructions with high fidelity due to its logic training.
- Supportive AI Interactions: Scenarios where a helpful and grounded AI persona is beneficial alongside strong reasoning.
For developers, a LoRA adapter and a Q4_K_M GGUF version are also available.