tannedbum/L3-Rhaenys-8B
tannedbum/L3-Rhaenys-8B is an 8 billion parameter language model, merged from Sao10K/L3-8B-Stheno-v3.2, Sao10K/L3-8B-Niitama-v1, and princeton-nlp/Llama-3-Instruct-8B-SimPO-v0.2 using the slerp method. This model is specifically configured for text completion tasks, particularly within roleplay and creative writing applications, as indicated by its optimized SillyTavern presets. It leverages the Llama-3 architecture and is designed for nuanced conversational generation.
Loading preview...
Overview
tannedbum/L3-Rhaenys-8B is an 8 billion parameter language model created by tannedbum through a series of merges using the mergekit tool. The model integrates components from three distinct base models: Sao10K/L3-8B-Stheno-v3.2, Sao10K/L3-8B-Niitama-v1, and princeton-nlp/Llama-3-Instruct-8B-SimPO-v0.2. The merging process utilized the slerp method, with specific parameter configurations applied to self-attention and MLP layers.
Key Capabilities
- Merged Architecture: Combines strengths from multiple Llama-3 based models for enhanced performance.
- Optimized for Text Completion: Includes specific SillyTavern presets for
temp,top_k,top_p,min_p,rep_pen,smooth_factor, andsmooth_curve, suggesting a focus on controlled and coherent text generation. - Instruct Mode Support: Designed to function effectively with instruct-based prompts, leveraging presets like those from Virt-io.
Good For
- Roleplay and Creative Writing: The provided SillyTavern presets and the model's origin suggest a strong suitability for interactive storytelling and character-driven narratives.
- Custom Text Generation: Developers looking for a merged Llama-3 variant with fine-tuned generation parameters for specific conversational or creative applications.