2025-02-19 19:10:03.459504+01 by Dan Lyke 1 comments
For work I'm playing around with hierarchical summarization techniques and ChatGPT4-o, and it's kind of astounding how different the output from separate runs with the same input text can be.
Like bullet points that are present in one query completely disappear on the second query, and of course the prioritization of such things is always very different.
[ related topics: Work, productivity and environment ]
comments in ascending chronological order (reverse):
#Comment Re: For work I'm playing around with made: 2025-02-19 19:56:08.312632+01 by: Dan Lyke
Copied over from the Mastodon conversation. re "then there’s the difference in answers and detail between paid and free versions.…"
I'm using the paid-for API for this stuff, so I'm not sure what all they'd be doing there.
Though the office response to the $200/month product on questions that we know some of the answers to is "there's no way it's actually doing the research it claims to be doing, because these other things qualify".
So, yeah, it's levels of credibility in bullshit generation, with the additional observation that we are more vested in believing something is not bullshit when we pay more for it.
LLM tiers are the audiophilia of the 2020s.
We will not edit your comments. However, we may delete your comments, or cause them to be hidden behind another link, if we feel they detract from the conversation. Commercial plugs are fine, if they are relevant to the conversation, and if you don't try to pretend to be a consumer. Annoying endorsements will be deleted if you're lucky, if you're not a whole bunch of people smarter and more articulate than you will ridicule you, and we will leave such ridicule in place.
Flutterby™ is a trademark claimed by
Dan Lyke for the web publications at www.flutterby.com and www.flutterby.net.