ave

8.2K posts

ave banner
ave

ave

@warnvod

lovable weirdo. kneady girl. mostly harmless. AS208590, DO4VE. she/her. also on @[email protected] and on the blue website as https://t.co/qDa5sckDpe

Hamburg 🍔 Katılım Kasım 2018
443 Takip Edilen1K Takipçiler
ave retweetledi
John Carmack
John Carmack@ID_AA_Carmack·
I know there is some overlap between open source and anti-AI activists, but I have a hard time reconciling it. My million+ open source LOC were always intended as a gift to the world. Yes, I would make arguments about how it would strengthen our communities, and the GPL would prevent outright exploitation by our competitors, but those were to allay fears of my partners to allow me to make the gift. AI training on the code magnifies the value of the gift. I am enthusiastic about it! Some people do look at open source as a tool for social change, career advancement, or reputation building, but those are all downstream of the gift.
Rich Whitehouse@DickWhitehouse

Genuinely devastating take to see from someone who popularized the GPL across so many communities. Fails to appreciate the social and cultural importance of the license.

English
175
325
3.4K
381.6K
ave
ave@warnvod·
@just_some_dev That's been there for like 5+ years now
English
0
0
2
35
ave
ave@warnvod·
@JohnKeough20 I'd also be curious if you have theories as to why a DOJ appointee spoke to "This Week in Worcester" (MA) rather than, say, CNN or Reuters.
English
1
0
0
29
ave
ave@warnvod·
@JohnKeough20 Hey John, a lot of people are understandably skeptical about such a large exclusive piece coming out of a small town newspaper. Did you do any validation about the claimed credentials of the sources, especially the DOJ appointee?
English
2
0
0
26
ave
ave@warnvod·
@Sosowski @Menezesworks As you said, it was trained in FP4, and IME quantized models perform worse than the equivalents trained directly in FP4, FP8 etc.
English
1
0
1
18
ave
ave@warnvod·
@tinybnuuy Any opinions on the amount of trans shooters in the last 10 years (💦)?
English
1
0
0
35
ave
ave@warnvod·
@Sosowski @Menezesworks Eh, I think the MXFP4ness of it is definitely nice as it allows people to run it without needing several thousand euros worth of GPUs (or thousands worth of RAM, thanks sama).
English
1
0
0
30
Sos Sosowski
Sos Sosowski@Sosowski·
@warnvod @Menezesworks Yeah that’s the problem. They probably don’t even have the rights to have the data. That’s why they will never release an open source model. OSS is cool but it’s not open source and also it sucks that it’s FP4
English
2
0
1
39
ave
ave@warnvod·
@Sosowski @Menezesworks Now the distinction isn't massive, but I think it's worth mentioning it's not exactly the same thing. Legally speaking, they likely cannot publish all of their training data because they don't have the right to do so.
English
1
0
0
37
ave
ave@warnvod·
@Sosowski @Menezesworks I said open weights, not open source. I don't think your winrar analogy is quite right though, close to it but not exactly. It is a binary, but you are free to modify them and change their behavior. You cannot legally do that with winrar.
English
1
0
0
38
etherret🐾
etherret🐾@witchof0x20·
I was thinking "what if you just poisoned the outputs of known distillers" but bad outputs are known to correlate with misalignment so this is likely to be evil in the long run. The best solution is to route to an aligned model with bad outputs:
etherret🐾 tweet media
English
6
3
71
2.9K
LucasMW
LucasMW@Menezesworks·
@Sosowski I think they released gpt2, but that is it
English
2
0
0
2.6K
ave
ave@warnvod·
@caninefaggot finanzamt... finanzamt it's been 1.5 months please decide on my tax returns finanzamt...
English
0
0
1
17
Sos Sosowski
Sos Sosowski@Sosowski·
@__alula @cursor_ai That's not about vibecoding, that's a fundamental part of how they operate. This tool has a non-zero chance of rm -rf'ing your entire home folder, and there is no option to disable that. Every time you run it it's like playing a russian roulette.
English
1
0
1
66
Alula
Alula@__alula·
why does @cursor_ai run commands "in sandbox" without my approval despite being configured *not to run any at all* without my approval 💀
Alula tweet mediaAlula tweet media
English
1
0
14
1.2K
ave
ave@warnvod·
@Daanniii6 to be fair after you handed in your exam you gave him an explosive apple that made his trashcan blow up
English
1
0
63
2.4K
ave
ave@warnvod·
@witchof0x20 That makes sense, I noticed it wasn't there on bsky today (where you have neither I think?) so I thought it was deliberate.
English
0
0
1
17
etherret🐾
etherret🐾@witchof0x20·
@warnvod I kept it originally because I thought the 🐾 wouldnt be immediately understood as a pun of equal meaning (with a mildly furry flair) but decided to just commit to being myself over legibility.
English
1
0
1
28
etherret🐾
etherret🐾@witchof0x20·
It occurs to me this is a testable-ish hypothesis. It's well known a lot of the "save 4o" guys talk like they've been brainjacked by 4o. When OAI shuts down the model, it will be interesting to observe how they talk (I suspect some are actual bots)
etherret🐾@witchof0x20

@uwukko either a lot of these posts are written by 4o directly or we need to start taking the guys on lesswrong talking about cognitohazardous parasitic ai way more seriously

English
1
0
6
240
ave
ave@warnvod·
@Noah_Stillhard @Ezkie_Music @uwukko i read through. you have no idea what you're talking about and this shit is borderline offensive tbh. people die of this you know?
English
0
0
3
27
wukko
wukko@uwukko·
i just saw a tweet from one of “keep 4o” psychos and holy fuck openai should’ve killed that model way earlier
English
83
156
4.8K
137.7K