DarkArtsForge/Paimon-24B-v3
DarkArtsForge/Paimon-24B-v3 is a 24 billion parameter language model based on the MistralForCausalLM architecture, created through a DELLA merge of multiple pre-trained models, with mistralai/Mistral-Small-24B-Instruct-2501 as its base. This model is specifically designed for generating narratives and roleplay content, including potentially violent and graphic erotic material, and operates with a 32768 token context length. Its development focused on refining content generation by carefully selecting merged models to avoid positivity bias from the base model.
Loading preview...
Paimon 24B v3: A Specialized Merge Model
Paimon 24B v3 is a 24 billion parameter language model developed by DarkArtsForge, built upon the MistralForCausalLM architecture. It is the result of a sophisticated merge process using the DELLA merge method, with mistralai/Mistral-Small-24B-Instruct-2501 serving as its foundational base model.
Key Characteristics
- Merge Architecture: Utilizes the DELLA merge method, combining over ten distinct 24B models to achieve its specialized capabilities.
- Content Specialization: Explicitly designed and optimized for generating detailed narratives and roleplay scenarios, including content that may be violent or graphically erotic.
- Context Length: Supports a substantial context window of 32768 tokens, enabling longer and more coherent interactions.
- Refined Merging: The v3 iteration involved careful selection and exclusion of certain merged models to prevent re-activation of the original base model's positivity bias, ensuring its intended output characteristics.
Ideal Use Cases
- Creative Writing: Generating complex and detailed fictional narratives.
- Roleplay Scenarios: Creating dynamic and immersive roleplaying experiences.
- Content Exploration: For users specifically seeking models capable of producing explicit or mature content, with appropriate system prompt adjustments.