add plain-fp8 (e4m3) training base (block linears fp8, ~54% size)

Upload krea2-turbo-snofs0.75-fp8-scaled.safetensors with huggingface_hub

Upload krea2-raw-snofs0.75-bf16.safetensors with huggingface_hub

Upload README.md with huggingface_hub

initial commit