Comment by echelon
> Here Llama is predicting one sentence at a time when fed the previous one, with 50% accuracy, for 42% of the book. That can easily be written off as fair use.
Is that fair use, or is that compression of the verbatim source?
> Here Llama is predicting one sentence at a time when fed the previous one, with 50% accuracy, for 42% of the book. That can easily be written off as fair use.
Is that fair use, or is that compression of the verbatim source?
It doesn't let you recover the text without knowing it in advance, so no.
You can't in particular iterate it sentence by sentence; you're unlikely to go past sentence 2 this way before it starts giving you back it's own ideas.
The whole thing is a sleigh of hand, basically. There's 42% of the book there, in tiny pieces, which you can only identify if you know what you're looking for. The model itself does not.