I very much agree with this. It is a real concern with our industry, we seem indeed to keep reinventing the wheel a lot. How do we stop forgetting? How do we move forward?
Looks very interesting, I guess I will switch some of my devices to using this and we'll see how it goes.
Or why you can't trust large language model for any fact or knowledge related tasks...
Interesting post with a good perspective on big data projects over time. It confirms that most people don't fall in the big data bucket, it's likely less than a percent of the projects which would qualify.
Early days but could become an interesting alternative to Lua for an embedded scripting language in some projects.
Very good essay on why we shouldn't look down on the Luddite. They had plenty of their questioning right and it's actually pervasive now. We use the term as libel only because back then they lost...
This is a big milestone reached for that project. Let's hope it'll drive adoption up.
It's limits and biases are well documented. But, what about the ideologies of the people behind those models? What can it tell us about their aims behind those models? Questions worth exploring in my opinion.
Interesting work, trying to get back to the source material used by a generative model. This is definitely necessary as well.
Good explanation about how sampling works. Does a good job explaining why it shines and where it is limited.
This is lesser known and probably should stay obscure... don't do that at home kids. Those alternative operators (more like tokens really) don't help with readability at all and make not much sense with UTF8 code bases.
Looks like a young but interesting tool to assess the power consumption of a service. There's been quite some work in this domain on the client side, not so much on the server side. This is welcome.
Too often forgotten. Data is indeed a mean to an end. It's not outright knowledge and will require work to be useful. It better be aligned with your needs if you want to use it for decision making.
Interesting business model. Could be a new path for at least some maintainers to be sustainably funded. Still a lot of unknowns though... probably worth keeping an eye on it.
Excellent conclusion to the recent turmoil around undefined behaviors. The way they are currently used as definitely a problem, they shouldn't be ignored in your code (although that's admittedly not that easy). There's hopefully a path forward.
A few interesting points in there. Too much hype and important points are glanced over, we'd all benefit from them being more actively explored.
An interesting but sometimes forgotten possibility for extending SQLite. Keep in mind this can lead to bad coupling between the software and the DB though which could carry interesting challenges around upgrades for instance.
OK, this is the best critique of the "Spotify Model" I've seen around. There's been plenty of unfair criticism thrown at this "model" (never aimed to be something you fully replicate though, hence the complaints I think). This one is properly balanced and doesn't just throw everything in the garbage bin, it takes the model bits by bits and try to highlight where the limits are. Very constructive.
Very nice summary of the architecture in the latest trend of transformer models. Long but comprehensive, a good way to start diving in the topic.
It seems this isn't necessary after all. At most if you like it you can put the year of creation of the copyrighted content, but the range and bumping it really isn't necessary.