Post B9Gi7syQv3TeSe1wAq by crunklord420@clubcyberia.co
(DIR) More posts by crunklord420@clubcyberia.co
(DIR) Post #B9Gi7syQv3TeSe1wAq by crunklord420@clubcyberia.co
2 likes, 1 repeats
CHUDBOUND 2 RELEASE IS IMMINENT
(DIR) Post #B9GiFZY11qlqWqL4im by s8n@posting.lolicon.rocks
0 likes, 0 repeats
@crunklord420 in the release announcement you should calculate how long it took to develop Chudbound 2 vs how long it has taken so far to develop Heartbound and declare that by the time it's finished we could already have Chudbound 36 or whatever
(DIR) Post #B9GiPZdnXLmziJoRoO by icedquinn@blob.cat
0 likes, 0 repeats
@s8n @crunklord420 i watched that guy "who worked at blizard for 7 years" code thinking i'd learn something. he writes about as badly as this new zealand guy who self-admittedly doesn't know what he's doing.:neocat_sweat:
(DIR) Post #B9GifuqaSavU7KaSsy by icedquinn@blob.cat
0 likes, 0 repeats
@phnt @crunklord420 @s8n yeah the new zealand guy was doing THIS EXACT SCUFF too.const e19 = {"what", "is a", "schema"}const e20 = {"fucking", "json", "please"}i asked him if he thought this should maybe be a json file and he just told me "they're both just text files" and i was like :blobcatgooglysoldering: you know what, have fun
(DIR) Post #B9Giq2Ql7VPeX5Khoe by s8n@posting.lolicon.rocks
0 likes, 0 repeats
@icedquinn @crunklord420 Maldavius? Why were you watching him? That nigger is a pervert who probably molests rodents, you didn't hear about all the controversy around him?
(DIR) Post #B9GisJjlkHOmmsJNWS by icedquinn@blob.cat
0 likes, 0 repeats
@phnt @crunklord420 @s8n the "that would just take longer" cope is golden considering he's using godot where loading json is a literal one liner
(DIR) Post #B9Gj1GDMDEFAT4PpYW by crunklord420@clubcyberia.co
1 likes, 1 repeats
@s8n it took about 2.5 months to make ChudBound 2. You can basically see my activity via my Claude usage chart.I started using Claude and work on CyberChud 2 basically at the same time. On a $20/month plan.
(DIR) Post #BA9lQBmyQTgxiLTfmK by s8n@posting.lolicon.rocks
0 likes, 0 repeats
@LazyLoungingLain @crunklord420 idk if that's really feasible right now. The AI needs to have access to search the internet for documentation and answer its questions and shit so it needs at least some agentic capabilities. Running an agent locally is like bending over and adopting the goatse pose and asking for some remote code execution
(DIR) Post #BAJU6UrXyMKUGCvaGO by crunklord420@clubcyberia.co
0 likes, 0 repeats
@LazyLoungingLain @s8n yes and it's kinda pointless, unless you have like a bunch of RTX 6000s and running a "real" model. You'd probably get further abusing free hosted LLMs.It's just very slow, and you're not getting anywhere close to the same number of parameters. It's just not viable to run this stuff economically on consumer hardware. The cost of buying 4x RTX 6000s so you can run GLM or whatever is huge compared to just paying per month. Only time and hardware market cycles will change this. It's going to take decades. We're back to the 1970s where your computer is a mainframe in another building.
(DIR) Post #BAJZUJaU2pTY0n7hdA by s8n@posting.lolicon.rocks
0 likes, 0 repeats
@crunklord420 @LazyLoungingLain I'm not so sure about that. The pace at which video models have been improving is mindblowing. It will only take one or two leaps like we just saw in video generation for local LLMs to be as good as Claude is right now.I really think the security issues are the real show stopper. The fact that you can't run the AI on the same machine you log into your bank from.