| Property | Value |
|---|---|
| Parameter Count | 70B |
| Base Model | LLaMA 3.3 |
| Merge Method | SCE (Selective Channel Ensemble) |
| Model URL | https://huggingface.co/zerofata/L3.3-GeneticLemonade-Unleashed-70B |
What is L3.3-GeneticLemonade-Unleashed-70B?
L3.3-GeneticLemonade-Unleashed-70B is an advanced language model created through a sophisticated merging process of multiple high-performance models. It's specifically designed for roleplay (RP) applications, striking a careful balance between creative writing capabilities and analytical intelligence. The model represents the culmination of extensive experimentation with merge techniques, inspired by the Nevoria series.
Implementation Details
The model utilizes a two-stage SCE merge process, first creating a base merge combining DeepSeek-R1-Distill, llama3.1-kartoffeldes, Llama-3.1-Swallow, and Negative_LLAMA. This was followed by a second merge incorporating EVA-LLaMA-3.33, Wayfarer-Large, nemotron-sunfall, and Hanami models.
- Recommended Temperature: 0.9-1.0
- MinP: 0.03-0.05
- Dry Run Settings: 0.8, 1.75, 4
- Available in 4.5bpw and 6bpw EXL2 quantizations
Core Capabilities
- Enhanced creative writing and roleplay performance
- Balanced intelligence and creativity
- Uncensored output generation
- Improved context understanding
- Compatible with LlamaTherapist instruction format
Frequently Asked Questions
Q: What makes this model unique?
This model's uniqueness lies in its carefully crafted merge process and optimization for roleplay scenarios while maintaining high intelligence. The selective merging of multiple specialized models creates a versatile system that excels in creative writing while maintaining coherent and contextually appropriate responses.
Q: What are the recommended use cases?
The model is primarily designed for roleplay applications and creative writing scenarios. It performs best in situations requiring a balance of creativity and logical coherence, making it suitable for interactive storytelling, character dialogue generation, and narrative development.