CoolFace
Modelpublic

mlx-community/RavenX-CyberAgent-Qwen3.6-35B-A3B-Opus-4.7-OpenMythos-Pentester-BugHunter-RATH-mlx-4bit-mtp-msq

sourceHugging Faceapache-2.0updated 3mo agoView on Hugging Face
7likes796downloads
Model Card

RavenX-CyberAgent-Qwen3.6-35B-A3B-Opus-4.7-OpenMythos-Pentester-BugHunter-RATH-mlx-mtp — MLX 4.5 BPW

Mixed-precision MLX quantization of `deadbydawn101/RavenX-CyberAgent-Qwen3.6-35B-A3B-Opus-4.7-OpenMythos-Pentester-BugHunter-RATH-mlx`, quantized with MLX Smart Quantize (MSQ) — my own sensitivity-based mixed-precision quantization method for Apple Silicon. It measures per-layer NMSE and assigns optimal bit widths automatically, combining architecture knowledge with measured data.

Details

  • —Type: Text (LLM)
  • —Average: 4.50 bits per weight
  • —Method: MLX Smart Quantize (MSQ)
  • —AWQ scaling: applied to 50 groups