Update Readme.md

[valentimarco]

New versions of llama-cpp-python needs a .gguf that can be created from old format by convert-llama-ggmlv3-to-gguf.py script in llama.cpp

[pieroit]

Good catch!
Let's wait a little for GGUF to stabilize (I see all GMML models are being converted on huggingface) and update also here

[pieroit]

@valentimarco I don't understand if "old" ggml models will still ve compatible or it is a breaking change

[valentimarco]

It's a breaking change, but they provide a script to convert old format.