| Property | Value |
|---|---|
| Parameter Count | 69B parameters |
| Model Type | GGUF Format |
| Context Length | 32k tokens |
| Format | Mistral |
What is miqu-1-70b?
miqu-1-70b is a large language model featuring 69 billion parameters, optimized for deployment through the GGUF format. It represents the first model in a potential series, implementing the Mistral architecture and instruction format for enhanced performance and versatility.
Implementation Details
The model utilizes a specific prompt format following the Mistral structure: "<s> [INST] QUERY [/INST] ANSWER</s>". It's configured with specialized RoPE (Rotary Position Embedding) settings optimized for a 32k token context window and high-frequency base, which shouldn't be modified for optimal performance.
- Optimized GGUF format for efficient deployment
- Specialized RoPE settings for extended context handling
- Recommended settings: Temperature 1.0, top_p 0.95
Core Capabilities
- Extended context processing with 32k token window
- Efficient inference through GGUF optimization
- Consistent performance with Mistral architecture
Frequently Asked Questions
Q: What makes this model unique?
The model combines a large parameter count (69B) with optimized GGUF format and extended context window, making it suitable for complex tasks while maintaining deployment efficiency.
Q: What are the recommended use cases?
While specific use cases aren't detailed, the model's architecture and size make it suitable for general-purpose language tasks, particularly those requiring extended context understanding and complex reasoning.