I never knew it was even possible to do line breaks in markdown any other way than double newline. Photon uses marked to parse markdown, I might just need to enable something in there.
I was about to explain why I probably wouldn't "fix" this, but I came up with a solution while explaining the problem lol. This'll be fixed with the next major release
The only web client that doesn't use JavaScript is mlmym. the modern slopware on the web where websites pump 5 MB of scripts straight to your browser is insanity, but JavaScript can also make websites much much nicer to use (routers that don't require full page reload).
Photon's Lemmy V1 compatibility is actually pretty much complete with a lot of technical improvements as well, I'm kinda hoping the Lemmy devs release Lemmy v1 soon as I don't really want to make a lemmy v0.x adapter
The situation is pretty confusing, but you're still gonna be able to side load unverified apps. The "advanced flow" that makes sideloading obnoxious to do is specifically for unverified apps, whereas verified apps (where devs have to fork up their ID to google) can be installed without any hassle.
You can still sideload apps. While the change is ridiculous, you still can sideload apps. They just make it significantly more painful to enable sideloading.
having llm output in your training set is irrecoverable
i used to think model collapse was an actual problem for LLMs as well, but it turns out that most popular models nowadays use intentionally synthetic data for things like reasoning traces and math. a lot of models (like gemini) also have subtle watermark patterns that let the trainers just filter out llm responses for factual data
I never knew it was even possible to do line breaks in markdown any other way than double newline. Photon uses
markedto parse markdown, I might just need to enable something in there.I was about to explain why I probably wouldn't "fix" this, but I came up with a solution while explaining the problem lol. This'll be fixed with the next major release
I can take it off your hands if you wanna clear up some space in your basement 😃
Sure, I'll add that. The share button on posts does already include "origin server" and "your server" buttons, though I need to add this to comments.
The only web client that doesn't use JavaScript is mlmym. the modern slopware on the web where websites pump 5 MB of scripts straight to your browser is insanity, but JavaScript can also make websites much much nicer to use (routers that don't require full page reload).
Photon's Lemmy V1 compatibility is actually pretty much complete with a lot of technical improvements as well, I'm kinda hoping the Lemmy devs release Lemmy v1 soon as I don't really want to make a lemmy v0.x adapter
it's not my job to moderate the fediverse, you won't need to worry about seeing any of this in photon
It's a 50/50 whether it lets me view it or not
They really need to add a way to view mod removed posts
There's probably some restrictions because I tried making an instance that had a user with a completely empty username and it didn't federate.
that mentality is one reason websites spam JavaScript today, because native features of the markup language were being added too slow
i do remember doing that a long time ago
MDN does, not my decision
The situation is pretty confusing, but you're still gonna be able to side load unverified apps. The "advanced flow" that makes sideloading obnoxious to do is specifically for unverified apps, whereas verified apps (where devs have to fork up their ID to google) can be installed without any hassle.
You can still sideload apps. While the change is ridiculous, you still can sideload apps. They just make it significantly more painful to enable sideloading.
I would use phtn.app because it lets you log into any instance, and it's also up-to-date. The lemmy.world version is over a year old
irrelevant to the post but this is about the first image with ultra HDR ive ever seen on lemmy.
i used to think model collapse was an actual problem for LLMs as well, but it turns out that most popular models nowadays use intentionally synthetic data for things like reasoning traces and math. a lot of models (like gemini) also have subtle watermark patterns that let the trainers just filter out llm responses for factual data
i doubt that Lemmy is being intentionally scraped by AI companies, otherwise it'd give their LLMs even more severe brain damage.
you saw LLMs being used in medicine almost 15 years ago? was it something else because the the transformer model was invented in 2017