LLMs get lost in multi-turn conversation

Post by westurner » Thu, May 15, 2025, 4:07 AM UTC

alganet wrote: post #30

morsecodist wrote: post #10

Earlier quoted context omitted.

Subjectively the "getting lost" feels totally different than human conversations. Once there is something bad in the context it seems almost impossible to get back on track. All subsequent responses become get a lot worse and it starts contradicting itself. It is possible that with more training this problem can be improved, but what is interesting to me isn't it's worse than humans in this way but that this sort of…

Contradictions are normal. Humans make them all the time. They're even easy to induce, due to the simplistic nature of our communication (lots of ambiguities, semantic disputes, etc). I don't see how that's a problem. Subjectivity is part of human communication.

Post by tsunamifury » Thu, May 15, 2025, 4:12 AM UTC

Post by olalonde » Thu, May 15, 2025, 4:17 AM UTC

unshavedyak wrote: post #16

Benjammer wrote: post #2

It's nice to see a paper that confirms what anyone who has practiced using LLM tools already knows very well, heuristically. Keeping your context clean matters, "conversations" are only a construct of product interfaces, they hurt the quality of responses from the LLM itself, and once your context is "poisoned" it will not recover, you need to start fresh with a new chat.

Has any interface implemented a .. history cleaning mechanism? Ie with every chat message focus on cleaning up dead ends in the conversation or irrelevant details. Like summation but organic for the topic at hand? Most history would remain, it wouldn’t try to summarize exactly, just prune and organize the history relative to the conversation path?

Post by kqr » Thu, May 15, 2025, 4:20 AM UTC

unshavedyak wrote: post #16

Benjammer wrote: post #2

It's nice to see a paper that confirms what anyone who has practiced using LLM tools already knows very well, heuristically. Keeping your context clean matters, "conversations" are only a construct of product interfaces, they hurt the quality of responses from the LLM itself, and once your context is "poisoned" it will not recover, you need to start fresh with a new chat.

Has any interface implemented a .. history cleaning mechanism? Ie with every chat message focus on cleaning up dead ends in the conversation or irrelevant details. Like summation but organic for the topic at hand? Most history would remain, it wouldn’t try to summarize exactly, just prune and organize the history relative to the conversation path?

Post by sitkack » Thu, May 15, 2025, 4:24 AM UTC

Terr_ wrote: post #22

Sharlin wrote: post #20

Earlier quoted context omitted.

This is a tired semantic argument that does not bring any insight into the discussion. A token-predictor could still be trained to predict the tokens “I’m not sure what you mean because of points x, y, and z; could you elaborate?”

It means if you want something resembling a self-introspective theory of mind, you need to arrange the overall document to cohere to documents where such things are/appear-to-be happening. This leads us to new questions: How can we characterize and identify real-world documents which fit? How can we determine what features may be significant, and which of those can be easily transplanted to our use-case?