| Property | Value |
|---|---|
| Author | Yoesph |
| Parameter Count | 24B |
| Base Model | arcee-ai/Arcee-Blitz |
| Model URL | Hugging Face |
| Merge Method | Model Stock |
What is Haphazard-v1.1-24b?
Haphazard-v1.1-24b is a sophisticated merged language model that combines multiple specialized models using the Model Stock merge method. Built on the Arcee-Blitz base model for general intelligence, it integrates three distinct models: Cydonia-24B-v2.1 for roleplay capabilities, Forgotten-Safeword-24B-V3.0 for uncensored content, and Dans-PersonalityEngine-V1.2.0-24b for enhanced prompt adherence.
Implementation Details
The model utilizes bfloat16 data type for efficient processing and memory usage. The merge was performed using mergekit, combining carefully selected models to create a balanced and capable language model.
- Base Model: arcee-ai/Arcee-Blitz for general intelligence foundation
- Merge Implementation: Model Stock methodology
- Data Type: bfloat16 for optimal performance
- Component Models: Three specialized models merged for comprehensive capabilities
Core Capabilities
- Enhanced roleplay capabilities from Cydonia-24B integration
- Improved prompt adherence through Dans-PersonalityEngine
- Comprehensive content generation abilities
- Balanced performance across multiple use cases
Frequently Asked Questions
Q: What makes this model unique?
This model's uniqueness lies in its strategic combination of specialized models, each contributing specific strengths: general intelligence from Arcee-Blitz, roleplay capabilities from Cydonia, and enhanced prompt handling from Dans-PersonalityEngine.
Q: What are the recommended use cases?
The model is well-suited for applications requiring strong prompt adherence, roleplay scenarios, and general language tasks. Its merged architecture makes it versatile across various use cases while maintaining specialized capabilities.