irrelevant to the post but this is about the first image with ultra HDR ive ever seen on lemmy.
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 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
Gnome seems to have amnesia whenever using sftp. disconnect once? say goodbye to ALL of your bookmarked network folders even after reconnecting.
i am going to crash out if ONE more website with an average of 4 monthly users puts on cloudflare "Under Attack Mode" and makes me sit through 15 seconds of verification before I can view their stupid webpage
I'll get to work on this around the end of May
It was down for about 6 hours yesterday. some bot farm spammed my cloudflare workers and used up all of my daily request quota within a day, which caused this to happen. after my usage limit reset it came back up. the bot farm seems to have stopped.
that'll show them, Google will be in shambles after some random app with like 1k downloads that makes them negative revenue threatens to REMOVE themselves from the PLAY STORE!!!!!
my problem isn't really about the topic at hand, it's moreso that the community has so many controversies constantly that nobody will really ever want to join. The user base will slowly dwindle away
view more: next ›
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