metadata
library_name: mlx
tags:
- mlx
- oq
- quantized
This quantization was uploaded on 2026-08-28 and replaces a previous version. If you downloaded this model before this date, please re-download for the updated weights.
Qwen3.8-4B-Distill-MLX-oQ4e-fp16-mtp
This model was quantized using oQ (oMLX v0.6.3) mixed-precision quantization.
Quantization details
- Model type: qwen3_5
- Bits: 4
- Group size: 64
- Format: MLX safetensors