ArcadaLabs/Ouro-2.6B-mlx-bf16

Remove unused Any import

Fix editorial content in code blocks, clarify excerpts vs verbatim output

Remove unused import

Fix broken GH link in Setup Notes, flag anomalous think tags in base output

Fix broken GitHub link, point to HuggingFace source model

Add measured peak memory for inference

Be more honest about benchmark limitations

Add benchmark prompt, outputs, and setup documentation

Upload unquantized bf16 MLX conversion of Ouro-2.6B

initial commit