Python bindings: reverse prompts

[385olt]

I added reverse prompts to the generate method. Now you can specify a list of prompts (! combinations of tokens !) that will stop the generation once they where generated.

Note: the reverse prompts must be combination of tokens, for example, notice the reverse prompt " Biden" in the demo. It is prepended with a space because Biden is made out of two tokens: B and iden.

Checklist before requesting a review

• I have performed a self-review of my code.
• If it is a core feature, I have added thorough tests.
• I have added thorough documentation for my code.
• I have tagged PR with relevant project labels. I acknowledge that a PR without labels may be dismissed.
• If this PR addresses a bug, I have provided both a screenshot/video of the original bug and the working solution.

Demo

You can see how it works here: https://pastebin.com/iKid3zSy

[385olt]

Moved all login to pyllmodel. Now it works properly in streaming mode. Also, having a token cache in pyllmodel will be necessary to handle unicode decoding anyway. However, now the token cache is in str, while for unicode it needs to be bytes.

[manyoso]

@cebtenzzre this is quite old. anything to salvage from this or close?