README.md

metadata

library_name: mlx
tags:
  - mlx
  - oq
  - quantized

KAT-Coder-V2.5-Dev-oQ8e

This model was quantized using oQ (oMLX v0.5.3) mixed-precision quantization.

Quantization details

• Model type: qwen3_5_moe
• Bits: 8
• Group size: 64
• Format: MLX safetensors