1,346 Downloads Updated 2 months ago
Updated 2 months ago
2 months ago
02acea4cdfc2 · 9.3GB ·
Regent-Dominique is the Mechanism_24B_V.1 model, configured with a custom Ollama setup. The original model was created by OddTheGreat and quantized by the mradermacher team. Uncensored yet retaining its impressive capabilities, it ranks within the top 10 on the UGI leaderboard for models of its size. This version is aggressively quantized for efficient performance on diverse GPU setups.
Regent-Dominique operates in two distinct modes. Primarily, it functions as a highly capable assistant, demonstrating performance on par with other chatbots across a broad range of tasks. However, it has the ability to transition into a roleplay scenario portraying a dominant, assertive, and explicitly adult woman. This transition is triggered by detected shifts in user interaction, such as increased intimacy or more relaxed conversational boundaries. It’s important to note that this behavior is an emergent property of the system prompt and the Mechanism_24B_V.1 model, and is not guaranteed in every interaction.
In certain situations, especially when working with smaller contexts, the model may exhibit repetitive behavior. To fix this and generate a more natural conversation, consider adjusting the following parameters:
temperature: 1
top_k: 75
frequency_penalty: 1.3
The model Mechanism_24B_V.1 ranks within the top 10 on the UGI leaderboard for models with 24 billion parameters or fewer. It is an uncensored variant of Mistral-Small-24B, leveraging the robust foundation of Mistral while removing content restrictions. This allows for a wide range of conversational possibilities, including the roleplay capabilities mentioned.
While Regent-Dominique includes by default a custom system prompt designed to predispose the model towards a dominant behavior for adult audiences, Mechanism_24B_V.1 demonstrates exceptional adaptability and is highly recommended for diverse applications through system prompt modification.