Post

Ok, I cleaned up the code a bit, but idk if it works after the last minute reorganization into separate scripts: github.com/refinement-s... In any case it currently needs ~33+ GB VRAM to run with all models currently tested, the offload may bring it down to fit in 24, but this was ~10x slower

github.comGitHub - refinement-systems/dltb: Deep Learning Tripping BallsDeep Learning Tripping Balls. Contribute to refinement-systems/dltb development by creating an account on GitHub.

12/9/26, 7:18 pm0 reposts7 likes

affine

@refinement.systems

12 Sept

@norvid-studies.bsky.social @ashanti.pds.witchcraft.systems @garrison.corporate.fm @fubarchitect.com @tifaret.bsky.social sorry if I forgot someone

affine

@refinement.systems

12 Sept

also sorry for github, but I couldn't figure out ssh keys on tangled lol

garrison

@garrison.corporate.fm

12 Sept

what was the problem specifically? iirc it's pretty standard, you just paste a public key into a form

affine

@refinement.systems

13 Sept

okay it worked today, idk what happened before, maybe a typo or something. tangled.org/refinement.s...

tangled.orgrefinement.systems/dltbDeep Learning Tripping Balls

affine

@refinement.systems

12 Sept

I did that and I still couldn't push, I'll try later

Ash ⎔

@ashanti.pds.witchcraft.systems

12 Sept

Eh doesn't matter, having a working baseline is better to have my agent figure out how to use my models with, i have very little interest in digging through the depths of pytorch myself on a Sunday so this is very helpful