Hacker News

Favorites Setup
Comment by trollbridge | original | What my dad taught me about AI coding in the 90s
[−]trollbridge · 2026-08-30 Sun 14:27 UTC · link

  LLMs are notoriously non-deterministic, and even at temperature zero you still can't predict exactly where the weights will take you next.
An LLM can be made to be completely deterministic. I use them in this mode so I can reproduce test cases. Of course it requires complete control over the model, etc. but this myth that a computer program is non-deterministic needs to end.

You can 100% predict where the weights “will take you” given a set of inputs.

[−]Folcon · 2026-08-30 Sun 14:33 UTC · link
> You can 100% predict where the weights “will take you” given a set of inputs.

Do you mean reproduce?

Sorry it's just if you are saying what your statement implying then either the model is very simple, or you've figured out something incredible

[−]techpression · 2026-08-30 Sun 14:35 UTC · link
You should publish, likely a Nobel price or Turing award waiting, and generational wealth at some tech giant.
[−]simonw · 2026-08-30 Sun 14:35 UTC · link
By "can't predict exactly where the weights will take you next" I meant with your brain. The blind chess analogy suggests you can predict, using your own thought process, the exact output of a prompt.
[−]shakna · 2026-08-30 Sun 14:45 UTC · link
Floating point matrix calculations are non-deterministic. You need to invent new hardware, that doesn't use floating point math, first. [0]

[0] https://arxiv.org/html/2506.09501

[−]mrob · 2026-08-30 Sun 15:29 UTC · link
>Floating point matrix calculations are non-deterministic.

This is not inherent to floating-point math. That actual (true) claim in the article is that different hardware and different hardware configurations produce different results. But deterministic inference is possible, e.g. llama.cpp on CPU is deterministic by default.

[−]feelamee · 2026-08-30 Sun 15:42 UTC · link
using which floating point standard? IEEE754 is totally deterministic
[−]andai · 2026-08-30 Sun 15:23 UTC · link
Could you give some examples?
[−]simonw · 2026-08-30 Sun 15:40 UTC · link
Can you provide steps to reproduce so I can see one of these deterministic LLMs running myself? API based or local models.