Document why MLX is not shipped: 2.9x slower than Core ML and non-finite output f3cedaa verified msluszniak commited on 9 days ago