News for hackers. Stuff that matters.

A comment on Can LLMs write better code if you keep asking them to "write better code"?. Parent comment

Re:Can LLMs write better code if you keep asking them to "wr... by ThrowawayR2 on Friday January 03, 2025 @11:15AM (#42586869)
We've entered the voodoo witch doctor phase of LLM usage: "Enter thee this arcane incantation along with thy question into the idol and, lo, the ineffable machine spirits wilt be appeased and deign to grant thee the information thou hast asked for."
[ Parent | on HN ]
Re:Can LLMs write better code if you keep asking them to "wr... by becquerel on Friday January 03, 2025 @11:37AM (#42587083)
This has been part of LLM usage since day 1, and I say that as an ardent fan of the tech. Let's not forget how much ink has been spilled over that fact that "think through this step by step" measurably improved/improves performance.
[ Parent | on HN ]
Re:Can LLMs write better code if you keep asking them to "wr... by RansomStark on Friday January 03, 2025 @02:41PM (#42588787)
> "think through this step by step"

Has always made sense to me, if you think how these models were trained.

My experience with great stackoverflow responses and detailed blog posts, they often contain "think through this step by step" or something very similar.

Intuitively adding that phrase should help the model narrow down the response content / formatting

[ Parent | on HN ]
Re:Can LLMs write better code if you keep asking them to "wr... by zahlman on Saturday January 04, 2025 @08:04AM (#42594460)
Then why don't they hard-code the interface to the model to pretend you included that in the prompt?
[ Parent | on HN ]
Re:Can LLMs write better code if you keep asking them to "wr... by Phlogistique on Saturday January 04, 2025 @09:53AM (#42595012)
...they do. It's called "o1".
[ Parent | on HN ]
Re:Can LLMs write better code if you keep asking them to "wr... by jmathai on Friday January 03, 2025 @12:13PM (#42587439)
We use the approaching of feeding mistakes from LLM generated code back to the LLM until it produces working code [1].

I might have to try some more aggressive prompting :).

[1] https://withlattice.com

[ Parent | on HN ]
Re:Can LLMs write better code if you keep asking them to "wr... by ThrowawayTestr on Friday January 03, 2025 @12:49PM (#42587810)
Praise the Omnissiah
[ Parent | on HN ]
Re:Can LLMs write better code if you keep asking them to "wr... by bigmutant on Friday January 03, 2025 @12:53PM (#42587857)
The Tech-Priests of Mars are calling
[ Parent | on HN ]
Re:Can LLMs write better code if you keep asking them to "wr... by manishader on Saturday January 04, 2025 @06:43AM (#42594143)
It is because the chance of the right answer goes down exponentially as the complexity of what is being asked goes up.

Asking a simpler question is not voodoo.

On the other hand, I think many people are trying various rain dances and believing it was a specific dance that was the cause when it happened to rain.

[ Parent | on HN ]