Ask HN: How do you avoid / get out of LLMs local minima?

3 points by d--b 19 days ago | 3 comments

When using LLMs, I have the impression that they often stick to an idea which they dont seem to be able to shake off, no matter how much arguing I can put in, and however strongly I feel about it.

What's your trick to get the LLM out of its own beliefs ?

I often find myself having to kill the entire context, which is obviously not the best...

Jimmc414 18 days ago |

This is conversational momentum due to the autoregressive nature of models. Each token is drawn from a probability distribution conditioned on the preceding tokens. The best method I know of working around this is to request and curate a markdown export package from a conversation that is used to prime a new conversation in a clean context.

skvark 18 days ago |

Having a separate planning / research phase helps with this. Make the LLM curate a plan by gathering internal and external context. Then execute the plan in another fresh session. Of course if the planning phase itself ends up in the local min then I would just start a new planning session with the learnings.

uberman 19 days ago |

I have found that in such situations all one needs to do is ask "are you sure x is y" atleast as far as most Claude models go. That usually results in and apology and and escape from your local min.