EigenLabs/Qwen3.6-35B-A3B-MLX-mixed-4bit
Different architecture vs Qwen/Qwen3.6-35B-A3B-FP8. Different architecture: only 331 of 64196/1477 tensors match by name and shape.
Different architecture vs Qwen/Qwen3.6-35B-A3B-FP8. Different architecture: only 331 of 64196/1477 tensors match by name and shape.