Mawdistical/RAWMAW-70B
Mawdistical/RAWMAW-70B is a 70 billion parameter instruction-tuned language model based on TheSkullery/Unnamed-Exp-70b-v0.7.A, developed by Mawdistical. This model is specifically fine-tuned for generating content with explicit themes, including mild to heavy violence, dark themes, NSFL, and heavy NSFW content, often with a male-leaning, anthro, and xeno-likeness focus. It is designed for roleplay scenarios that may involve grey-line consent and passive negative biases, making it suitable for specific narrative and character-driven applications requiring mature and intense subject matter.
Loading preview...
Mawdistical/RAWMAW-70B Overview
RAWMAW-70B is a 70 billion parameter language model developed by Mawdistical, built upon the TheSkullery/Unnamed-Exp-70b-v0.7.A base model. It is specifically designed for generating explicit and mature content, with a strong focus on specific narrative themes and character interactions.
Key Characteristics & Training
- Base Model: TheSkullery/Unnamed-Exp-70b-v0.7.A.
- Tokenizer: Utilizes a LLAMA tokenizer.
- Training Dataset: Composed of 2500 unique set scenarios from a pool of 40,000, with a hybrid organic-synthetic mix.
- Content Focus: The dataset includes entries with:
- Mild & Heavy Violence
- Dark Themes
- Mild to Extensive NSFL (Non-lethal but intended to be unsettling)
- Heavy NSFW, triggered by user action or passively within roleplay.
- Modern City and Dystopia Scenes.
- Grey Line Consent Scenes, where consent may be implicit or action-based, potentially leading to possessiveness.
- Thematic Leaning: Primarily male-leaning, anthro, and xeno-likeness, with a passive negative bias and mild narcissistic tendencies.
- Token Count: 13.1 million tokens, calculated solely on outputs.
Recommended Usage
RAWMAW-70B is intended for use in scenarios requiring explicit and mature content generation, particularly for roleplay and narrative creation that aligns with its specific thematic focus. Users are advised to experiment with a temperature range of 0.6-0.9 to find their preferred output style.
Explicit Content Warning: This model is designed to generate explicit content, including themes of violence, dark subjects, and NSFW material. Users should be aware of and comfortable with these themes before deployment.