Converted locally from alphaZimuth/Hy-MT2-30B-A3B-APEX-GGUF/Hy-MT2-30B-A3B-APEX-Imatrix-I-Nano.gguf for oMLX/MLX inference.

The source GGUF mixed precision policy is preserved by mapping IQ2_XXS/IQ2_S/Q3_K/Q4_K/Q5_K/Q6_K tensors to MLX affine 2/3/4/5/6-bit modules with group size 64. Quantization parameters and ordinary floating weights use F16; expert routing biases remain F32.

Source tensor types: {'F32': 287, 'IQ2_S': 30, 'IQ2_XXS': 60, 'Q3_K': 163, 'Q4_K': 114, 'Q5_K': 27, 'Q6_K': 85}.