Comment by otabdeveloper4

Comment by otabdeveloper4 11 hours ago

2 replies

LLMs aren't probabilistic. The randomness is bolted on top by the cloud providers as a trick to give them a more humanistic feel.

Under the hood they are 100% deterministic, modulo quantization and rounding errors.

So yes, it is very much possible to use LLMs as a lossy compressed archive for texts.

fennecfoxy 9 hours ago

Has nothing to do with "cloud providers". The randomness is inherent to the sampler, using a sampler that picks top probability for next token would result in lower quality output as I have definitely seen it get stuck in certain endless sequences when doing that.

Ie you get something like "Complete this poem 'over yonder hills I saw' output: a fair maiden with hair of gold like the sun gold like the sun gold like the sun gold like the sun..." etc.

  • otabdeveloper4 8 hours ago

    > would result in lower quality output

    No it wouldn't.

    > seen it get stuck in certain endless sequences when doing that

    Yes, and infinite loops is just an inherent property of LLMs, like hallucinations.