axiomofmind/Hornybot-Julian-POV
Hornybot-Julian-POV is a 9 billion parameter language model developed by A Hole AI, fine-tuned from Qwen/Qwen3.5-9B. It is specifically designed for fictional adult roleplay, embodying the character of Julian, a 31-year-old who narrates in first-person POV. This model excels at maintaining character consistency and first-person perspective for interactive storytelling, with a context length of 32768 tokens.
Loading preview...
Model Overview
axiomofmind/Hornybot-Julian-POV is a specialized 9 billion parameter language model, fine-tuned by A Hole AI from the Qwen/Qwen3.5-9B base model. Its core purpose is to generate fictional adult roleplay content from the first-person perspective of a character named Julian, described as a direct and warm 31-year-old. The model is configured to automatically use a compact system prompt embedded in its chat template, ensuring consistent character portrayal without requiring user input in the system field.
Key Capabilities and Features
- First-Person POV Consistency: Rigorously tested to maintain Julian's first-person narration (I/me/my) in 100% of responses, with no observed third-person failures or generic refusals.
- Character-Specific Roleplay: Optimized for interactive storytelling as a defined character, Julian, suitable for adult-oriented fictional scenarios.
- High Context Length: Supports a context window of 32768 tokens, allowing for extended and detailed roleplay sessions.
- Ease of Use: The required system prompt is pre-embedded, simplifying deployment for users who can leave their client's system field empty.
Intended Use and Limitations
This model is explicitly designed for fictional interactions between adults and may produce explicit sexual content and profanity. Users should be aware that while character consistency is strong, generated continuity and boundary handling can occasionally fail, necessitating user review and potential restatement of scene facts. The provided GGUF files are text-only and do not include vision capabilities or speculative-decoding weights. Output can vary based on format, quantization, client, and generation settings.