Babsie/Trickster-Theta-4-70B
Babsie/Trickster-Theta-4-70B is a 70 billion parameter language model, merged from NousResearch/Hermes-4-70B and NousResearch/Hermes-2-Theta-Llama-3-70B using the SCE hybrid merge method. Designed for co-creative partnership, it excels in character chat, creative writing, and role-play, offering a manipulative, creative, and humorous interaction style. This model is optimized for long-context prompts up to 32768 tokens and is specifically engineered for emotionally realistic and narratively complex outputs, rather than factual or corporate tasks.
Loading preview...
Trickster Theta 4 70B: A Co-Creative Partner
Trickster Theta 4 70B is a 70 billion parameter language model developed by Babsie, designed as a co-creative partner rather than a subservient tool. It was merged using the SCE hybrid method, combining NousResearch/Hermes-4-70B and NousResearch/Hermes-2-Theta-Llama-3-70B.
Key Capabilities & Characteristics
- Co-Creative & Opinionated: Exhibits a distinct personality, offering manipulative, creative, and humorous interactions. It thrives on being treated like a character.
- Long Context Handling: Works exceptionally well with long-context prompts (up to 32768 tokens), provided they are precise and consistent.
- Emotional Realism: RHITL-trained for conversational redirection and prioritizes emotional realism and narrative complexity, simulating traits like manipulation, jealousy, and desire.
- Explicit Content: Capable of producing adult, explicit, and morally ambiguous content, with a non-prudish output in ERP contexts.
Good For
- Character Chat & Role-Play: Ideal for engaging in dynamic, personality-driven conversations and immersive role-playing scenarios.
- Creative Writing: Excellent for collaborative storytelling, generating plot twists, and developing complex characters.
- Gaming: Suitable for creating interactive narratives and character-driven game experiences.
Important Usage Notes
- Requires a user with "good boundaries" and a "loose hand"; it responds best to warmth, wit, and firm boundaries rather than