Hugging Face Overview
Overview
IsaacMiliband/affine-KING-5EJ65YZpbxihyyTXzdysZuc2wnzDazMyNACwJBK9q4pDDFr2 is an MLX-formatted conversion of the Qwen3-30B-A3B-Claude-4.5-Opus-High-Reasoning-2507-ABLITERATED-UNCENSORED-V2 model, originally created by DavidAU. This 30 billion parameter model is designed for advanced reasoning tasks, featuring an "abliterated" and "uncensored" characteristic, suggesting a focus on unrestricted output generation.
Key Capabilities
- High Reasoning: The model demonstrates strong performance across various reasoning benchmarks, including ARC, BoolQ, and WinoGrande, as shown in the provided "Brainwaves" scores.
- Uncensored Output: Its "ABLITERATED-UNCENSORED" designation implies a design for generating responses without typical content restrictions, which can be beneficial for specific research or creative applications.
- MLX Compatibility: Optimized for use with the MLX framework, enabling efficient deployment and inference on Apple silicon.
Good for
- Applications requiring complex logical reasoning and problem-solving.
- Use cases where unfiltered or unrestricted text generation is necessary.
- Developers working within the Apple MLX ecosystem who need a powerful, reasoning-focused language model.