SkySyrup

joined 2 years ago
MODERATOR OF
[–] SkySyrup 8 points 1 year ago (1 children)

my meme now

[–] SkySyrup 9 points 1 year ago (1 children)

I mean I agree with most of them but bri’ish?? Never!

[–] SkySyrup 8 points 1 year ago* (last edited 1 year ago)

This looks exactly like an image that would accompany an SCP article lol

[–] SkySyrup 6 points 1 year ago* (last edited 1 year ago)

Sure! You’ll probably want to look at train-text-from-scratch in the llama.cpp project, it runs on pure CPU. The (admittedly little docs) should help, otherwise ChatGPT is a good help if you show it the code. NanoGPT is fine too.

For dataset, maybe you could train on French Wikipedia, or scrape from a French story site or fan fiction or whatever. Wikipedia is probably easiest, since they provide downloadable offline versions that are only a couple gigs.

[–] SkySyrup 10 points 1 year ago
[–] SkySyrup 0 points 1 year ago

A simple way would be to load the comments themselves, and then check for blocked users. But this would basically ddos the instance servers and would be extremely janky lol

or I’m being me and missing something obvious :/

[–] SkySyrup 3 points 1 year ago

Thank you so much! Have a good break!

[–] SkySyrup 7 points 1 year ago

tech support?? that’s going a bit far I think

ok but if everyone would accept me when coming out I think it would be worth it..

[–] SkySyrup 18 points 1 year ago

kid called EU anticompetitive laws:

[–] SkySyrup 9 points 1 year ago* (last edited 1 year ago)

The technology of compression a diffusion model would have to achieve to realistically (not too lossily) store “the training data” would be more valuable than the entirety of the machine learning field right now.

They do not “compress” images.

[–] SkySyrup 6 points 1 year ago

can I interest you in eepy mode Tuesday?

[–] SkySyrup 2 points 1 year ago (1 children)

Are you using SDXL? If you are, you need to set the resolution to 1024x1024

view more: ‹ prev next ›