MN-Violet-Lotus-12B
| Property | Value |
|---|---|
| Parameter Count | 12.2B |
| License | CC-BY-4.0 |
| Base Model | Mistral-Nemo-Instruct-2407 |
| Paper | Model Stock Paper |
What is MN-Violet-Lotus-12B?
MN-Violet-Lotus-12B is an advanced language model specifically designed for creative writing and roleplaying applications. Built using the Model Stock merge method, it combines multiple specialized models to achieve superior emotional intelligence and storytelling capabilities. The model achieved an impressive 80.00 score on local EQ benchmarks with 100% parsing at 8-bit quantization.
Implementation Details
The model was created through a sophisticated merge of multiple pre-trained models including Violet_Twilight-v0.2, Lumimaid-v0.2-12B, Mahou-1.5-mistral-nemo-12B, and MN-12B-Lyra-v4. The merge process utilized mergekit and employed a strategic approach where models were combined using both SLERP and Model Stock methods, with Mistral-Nemo-Instruct-2407 serving as an anchor point.
- Implements BF16 tensor type for optimal performance
- Uses Model Stock merge methodology for enhanced weight distribution
- Incorporates specialized roleplaying and creative writing capabilities
Core Capabilities
- High emotional intelligence with verified 80.00 EQ score
- Advanced character personality adherence
- Sophisticated creative writing and storytelling
- Balanced output length control through Mahou integration
Frequently Asked Questions
Q: What makes this model unique?
The model's distinctive feature is its combination of high emotional intelligence with robust creative writing capabilities, achieved through a carefully orchestrated merge of specialized models. The impressive 80.00 EQ score sets it apart from similar models in the field.
Q: What are the recommended use cases?
This model excels in creative writing, roleplaying scenarios, and applications requiring strong character personality management. It's particularly well-suited for interactive storytelling and narrative generation where emotional intelligence is crucial.
