Real-time word-by-word stream as llama.cpp generates it - is it possible?

[ai2p]

So that won't be necessary to wait till the whole answer generated...

[rabidcopy]

Run with --stream.