Run DFloat11 losslessly compressed BF16 checkpoints on Apple Silicon with MLX and a Metal decode kernel; today that runs FLUX.1 on a 32 GB Mac with weights about 30% smaller
python macos flux machine-learning metal inference huffman-coding image-generation diffusion mlx model-compression memory-optimization lossless-compression bfloat16 apple-silicon mflux weight-compression dfloat11
-
Updated
Oct 9, 2026 - Python