Is there a consensus among #AI experts about whether #LLM models can reliably summarize specific text without introducing weird extraneous information?
My assumption has been that the models can't really do this, and I can't imagine a way that they could reliably avoid omitting important details or even add information, but I'm not sure I've seen much discussion of this aspect of what they can and can't do.
I was looking at this Sass (SCSS) compiler, written in #PHP, and I noticed something very odd.
Under “requires (dev)," it requires two packages, sass/sass-spec and thoughtbot/bourbon, both of which appear to be empty packages, containing only a composer.json file, which has no dependencies.
What’s the purpose of these packages? They otherwise appear suspicious, to me, but I can’t see that they're doing anything nefarious right now—they just appear pointless.
It looks like some kind of namespace hijack. The packages look like they were added to the project several years ago, but this guy has published new empty ones more recently. He's got a bunch of other repositories doing the same. Not sure what his endgame is, though.
Who else has trouble answering the “Occupation” field on forms, when it’s a predetermined list of options, with no “Other” option?
I guess the closest thing, for me, would be “Information Technology, Systems,” but I’ve spent much of my adult life explaining to others that I’m not in IT, so it feels wrong to select that.
I often got hit on this throughout my career as a software developer - it's weird that it wouldn't appear on many of these lists.
It isn't limited to our profession. My wife does meta-analysis of drugs and therapeutics, and manages the publication of this information. She usually gets lumped into "Health" or "Publishing" depending on whose list she gets to pick from.
I had high hopes that ViewCrafter would be a great development/design tool when I started working on it over five years ago. I struggled getting product-market fit, and I completely lost momentum on it in 2020 while I was occupied with family things, and never really returned to the project in earnest.
Today, I finally terminated the instance that was running the app.
Hi #BoardGames Mastodon, help me?
I'm trying to play #Décorum, which has a great conceit, but it looks like the two sets of conditions that we have to achieve are mutually exclusive. Am I getting something wrong?
:(
The maker, #Floodgate, is only accessible via Facebook and Twitter and "BGG" whatever that is - there's no link to that from their website.
At #DDDAdelaide session on burnout and recognising, I’ve ' been through this. Multiple times. My inability to recognise the problem, symptoms and triggers is pretty staggering. I’ve come out the other side each time, but not without consequences. Man I need to do a deep retrospective on my mental health and relationship to work.
That session was an eye opener - I kind of knew I was feeling #BurnOut this year, and have taken steps, and that was why I was motivated to attend the talk.
The studies indicating burnout symptoms can linger for 7 years hit hard - as I also went through burnout 5 years ago, and I realise that's probably been an ongoing factor in my professional and personal life since then.
Darth Sidious and Darth Vader imply the existence of Darth Fernal, Darth Structor, Darth Solent, Darth Tact, Darth Flammable, Darth Grown, and Darth Ternal Combustion Engine.