Archaeo-12B
| Property | Value |
|---|---|
| Parameter Count | 12 Billion |
| Model Type | Language Model |
| Architecture | SLERP Merge of Rei-12B and Francois-Huali-12B |
| Format | ChatML |
| Available Formats | GGUF, EXL2 |
What is Archaeo-12B?
Archaeo-12B is a sophisticated language model created by Delta-Vector, specifically designed for roleplay and creative writing applications. The model represents a careful merge of two foundational models - Rei-12B and Francois-Huali-12B - using SLERP (Spherical Linear Interpolation) methodology with a t-value of 0.2.
Implementation Details
The model employs ChatML formatting for conversation structure and is implemented using MergeKit configuration. It's available in both GGUF and EXL2 formats, making it accessible for various deployment scenarios. The merge was performed using bfloat16 dtype, ensuring efficient memory usage while maintaining performance.
- SLERP merge methodology with t=0.2
- ChatML conversation formatting
- Multiple quantization options (GGUF and EXL2)
- Base model: Rei-12B
Core Capabilities
- Specialized in roleplay and creative writing tasks
- Structured conversation handling through ChatML
- Optimized for creative content generation
- Enhanced narrative capabilities through model merging
Frequently Asked Questions
Q: What makes this model unique?
Archaeo-12B's uniqueness lies in its specialized merge of Rei-12B and Francois-Huali-12B models, specifically optimized for creative writing and roleplay scenarios. The careful SLERP merge with a 0.2 t-value creates a balanced model that leverages the strengths of both parent models.
Q: What are the recommended use cases?
The model is particularly well-suited for creative writing, storytelling, roleplaying scenarios, and narrative generation. Its ChatML formatting makes it ideal for interactive storytelling and conversation-based applications.
