FlareRebellion/DarkHazard-v1.2-24b
DarkHazard-v1.2-24b by FlareRebellion is a 24 billion parameter merged language model, built upon the arcee-ai/Arcee-Blitz base using the Model Stock method. This model integrates components from Eurydice-24b-v2, Broken-Tutu-24B, Cydonia-24B-v2.1, and Dans-PersonalityEngine-V1.2.0-24b, focusing on enhanced storytelling, uncensored content generation, NSFW capabilities, and strong prompt adherence. With a 32768 token context length, it is designed for applications requiring nuanced and unrestricted text generation.
Loading preview...
DarkHazard-v1.2-24b: A Merged Language Model
DarkHazard-v1.2-24b is a 24 billion parameter language model developed by FlareRebellion, created through a sophisticated merge of several pre-trained models. Utilizing the Model Stock merge method, it builds upon the arcee-ai/Arcee-Blitz base model.
Key Capabilities & Features
This model is a composite of four distinct models, each contributing specialized characteristics:
- aixonlab/Eurydice-24b-v2: Enhances storytelling and role-playing capabilities.
- TheDrummer/Cydonia-24B-v2.1: Contributes to uncensored content generation.
- ReadyArt/Broken-Tutu-24B: Further strengthens uncensored and NSFW content generation, incorporating aspects of Cydonia.
- PocketDoc/Dans-PersonalityEngine-V1.2.0-24b: Improves prompt adherence, ensuring the model follows instructions closely.
What Makes This Model Different?
Unlike many general-purpose LLMs, DarkHazard-v1.2-24b is specifically engineered through its unique merge configuration to excel in areas often restricted or less developed in other models. Its focus on combining storytelling prowess with uncensored and NSFW content generation, alongside robust prompt adherence, positions it for use cases requiring creative freedom and precise control over generated text, particularly in narrative and character-driven applications. The model's 32768 token context length supports extended and complex interactions.