Curious what people think are the ideal 4-bit quantization types on MLX These quants seem to be the most popular, at least for Gemma4 and Qwen3.6: - OptiQ 4bit (mlx-community/Qwen3.6-27B-OptiQ-4bit) - Unsloth dynamic 2.0 MLX (unsloth/Qwen3.6-27B-UD-MLX-4bit) - oQ (Jundot/Qwen3.6-27B-oQ4) - DWQ (can'
