Kunoichi-7B
| Property | Value |
|---|---|
| Parameter Count | 7.24B |
| Model Type | Text Generation, Transformers |
| License | CC-BY-NC-4.0 |
| Tensor Type | BF16 |
| Context Window | 8K (expandable to 16K) |
What is Kunoichi-7B?
Kunoichi-7B is an advanced language model created through a SLERP merger between Silicon-Maid-7B and an unreleased "Ninja-7B" model. It's specifically designed to excel in both general-purpose tasks and roleplay scenarios, achieving remarkable benchmark scores that rival much larger models.
Implementation Details
The model leverages the Mistral architecture and can be used with an 8K context window, with experimental support for up to 16K using NTK RoPE alpha of 2.6. It supports multiple prompt formats, including Alpaca and SillyTavern configurations.
- Achieves 8.14 on MT-Bench, surpassing many commercial models
- Scores 64.9 on MMLU, comparable to Starling-7B
- Shows strong performance on EQ-Bench (44.32) and Logic Test (0.58)
- Compatible with various prompt templates and configurations
Core Capabilities
- General-purpose text generation and roleplay
- Strong reasoning and analytical capabilities
- Enhanced context understanding and response coherence
- Flexible deployment options with different context windows
Frequently Asked Questions
Q: What makes this model unique?
Kunoichi-7B stands out for its exceptional balance between intelligence and roleplay capabilities, achieving benchmark scores that compete with much larger models while maintaining strong character adherence in RP scenarios.
Q: What are the recommended use cases?
The model excels in general text generation tasks, roleplay scenarios, and applications requiring strong reasoning capabilities. It's particularly well-suited for SillyTavern implementations and can be used with various prompt templates.
