Reposted by IskaDoll @dollspace.gay · 3hIf openssh which basically the entire internet relies on doesnt warrant mythos access then what does? Whats the criteria? Why isnt that criteria transparent? 512920
Reposted by IskaFinfet @finfet.sh · 14hpoorly working gaze tracking on steam frame. i've had it for less than a day. god i love mostly open hardware/software 1282
Reposted by Iskacyber professional @permadeath.com · 5hcome on man, it's not even 9 AM, at least let me caffeinate first 0484
Iska @iskamag.com · 4hWhy can't he finish his fish game like a normal person instead of waiting for a god model? 020
Iska @iskamag.com · 4hremember 4-8 months ago when oai/ant didn't have a quarter-decent model that didn't cost $15+ 110
Iska @iskamag.com · 5hIt still happens too often recently, getting blocked or banned for various disagreements. 240
Iska @iskamag.com · 5hShould honestly get better at Not Making Others Mad. Want to get along with everyone but it's hard... 1150
Iska @iskamag.com · 5hA friend was questioning AI usefulness, that a chatbot could save lives. Gave this post in reply, and got instantly blocked 🙃 (OP disabled quotes, sorry) 2352
Iska @iskamag.com · 7hThe skate + mw2 + minecraft guy said he decomped those 3 games and just prompted for them to be mashed together, probably being on the same engine 140
Reposted by IskaRayven (but more queer) @rayventhegm.bsky.social · 13hTIL there's still a thriving Spore community. www.spore.com/sporepedia#q... 121
Reposted by Iskamr. TIM @timkellogg.me · 29/09/2026NanoGPT pretraining runs now take 39.9 seconds for a 124M model(!!) now, 124M is *tiny* so it might not seem relevant. But much of the gains in LLM pretraining are data quality and a great way to find quality data is to train on it & measure the lift 3928
Iska @iskamag.com · 10hyeah charm usage is very decent. Never hit a burst limit with however many subagents 110
