Endevor/InfinityRP-v1-7B: An Experimental Roleplay Model
Endevor/InfinityRP-v1-7B is an experimental 7 billion parameter model developed by Endevor, primarily designed to address common issues like character hallucination and over-eagerness in roleplaying interactions. The model's core philosophy is to ensure the AI character ({{char}}) consistently waits for the user's ({{user}}) response and avoids making decisions or actions on behalf of the user, fostering a more collaborative and "cozy" roleplay environment.
Key Capabilities & Features
- Hallucination Mitigation: Specifically engineered to reduce instances of the AI character hallucinating or taking control of the narrative.
- User-Centric Interaction: Prioritizes waiting for user input, preventing the AI from prematurely advancing the story or deciding for the user.
- Structured Roleplay Formatting: Supports distinct formatting conventions for enhanced clarity in roleplay:
- Character dialogs are enclosed in double quotes (e.g.,
"Hey, Anon..."). - Narration can use asterisks or a default novel format.
- Character internal thoughts are wrapped with backticks (e.g.,
`My heart skips a beat!`), which can be encouraged via system prompts.
- Response Length Control: Accepts various response length parameters including
tiny, short, medium, long, and huge. - Alpaca Prompt Format: Utilizes the Extended Alpaca prompt format for instructions.
Ideal Use Cases
- Interactive Storytelling: Excellent for users seeking a more controlled and less intrusive AI partner in creative writing and collaborative storytelling.
- Character-Focused Roleplay: Suited for scenarios where maintaining strict character boundaries and user agency is paramount.
- Experimental AI Development: Useful for developers interested in exploring models designed with specific behavioral constraints for improved user experience in conversational AI.