allura-org/G2-9B-Aletheia-v1: Enhanced Narrative and Roleplay
Developed by Auri, G2-9B-Aletheia-v1 is a 9 billion parameter language model based on the Gemma-2-9B architecture, featuring a 16384-token context window. This model is a strategic merge of two distinct components: the G2-9B-Sugarquill-v0 model and a Gemma-2-9B-It-SPPO-Iter3 base model with an applied AuriAetherwiing/sunfall-g2-lora. The primary goal of this merge was to combine Sugarquill's novel-like writing style with the improved roleplay performance and steerability offered by the Sunfall component, without introducing synthetic writing patterns.
Key Capabilities
- Enhanced Writing Style: Produces fresh, novel-like prose suitable for creative writing and story generation.
- Improved Roleplay (RP) Performance: Designed to be more steerable and effective in interactive narrative scenarios.
- Character Card Adherence: Demonstrates good adherence to character descriptions and guidelines provided.
- System Prompt Following: Capable of accurately following instructions and constraints given in system prompts.
- Gemma Instruct Formatting: Responds using the standard Gemma instruct format.
Good For
- Creative Storywriting: Excels in generating engaging and coherent narratives.
- Interactive Roleplay: Ideal for applications requiring dynamic character interaction and plot progression.
- Content Generation: Suitable for tasks where a specific writing style and adherence to prompts are crucial.