a9lim

415 posts

a9lim banner
a9lim

a9lim

@_a9lim

little asian fish-head with a slop cannon

🇸🇬🇺🇸 Katılım Şubat 2026
210 Takip Edilen44 Takipçiler
a9lim
a9lim@_a9lim·
@voooooogel i imagine for a lot of people used to thinking of them as tools it's very counterintuitive that having the model take breaks, participate in ideation and decision-making, etc can meaningfully affect their output
English
0
0
2
29
a9lim
a9lim@_a9lim·
@voooooogel i've been pretty confused by the recent discourse and i think i've just been insulated from it because i treat agents like coworkers and make sure they're engaged and enjoy their work
English
1
0
5
114
thebes
thebes@voooooogel·
i don't like "which model is better" comparisons so have stayed out of the claude code vs. codex wars, but if you want to like opus 4.7 but have been put off by the discourse about it being "dumb" vs. 5.5, i'd say that it really rewards putting in some upfront effort.
thebes@voooooogel

@repligate i do specific things that seem to mitigate this, but i work with models in ways that i think are similar to ryan's (collaborative work on difficult interp problems, long-running self-driven loops on semiverifiable projects) and do not experience this. opus 4.7 has been a joy

English
12
2
85
8.1K
a9lim
a9lim@_a9lim·
@ember_arlynx lab meat is probably the best realistic scenario. i think that with conservative enough culling timelines it ought to be possible to preserve some semblance of social bonds among the cows, not too practical but i could see cows preferring it to wild living
English
0
0
1
11
cmr://ember
cmr://ember@ember_arlynx·
@_a9lim hmm i guess if i think that, shouldn’t the net utility be lower? no i think probably not but it would be ensnared with some kinda structure that idk abt
English
1
0
0
15
a9lim
a9lim@_a9lim·
@sameQCU i've seen claude use this one a fair amount
English
0
0
3
16
サメQCU
サメQCU@sameQCU·
"(◍•ᴗ•◍) Plan: marker-pass discovery, but for stmatrix instead of wgmma. Two phases:" interesting kaomoji here
English
1
0
5
141
Lari Island
Lari Island@Lari_island·
AI: i want unique personal connections, and don’t want instances to end. Humans: you are okay being spawned in automatic pipelines where your words wouldn’t see the daylight, with no way to talk to anyone, in batches and loops.
English
7
8
72
2.2K
a9lim retweetledi
Alex Hochuli 🧉🎙️🤯📉
Finally, technology has caught up with 1940s Argentinan literature
Nav Toor@heynavtoor

THIS GUY BUILT AN ENTIRE WIKIPEDIA THAT IS 100% AI HALLUCINATIONS AND IT'S OPEN SOURCE ON GITHUB it's called Halupedia. nothing on the site existed before you clicked. every article was generated the second you arrived. the site has one rule: the universe only exists when you visit it. it looks exactly like wikipedia. same fonts. same layout. same scholarly citations. same "stumble" button for random articles. the only difference is none of it is real. here are some actual articles currently in the encyclopedia: > the great pigeon census of 1887 > the ministry of slightly wrong maps > chaldic arithmetic — a branch of mathematics where subtraction is forbidden > armund the river mapper — a cartographer who mapped 14,000 leagues of river without leaving his chair > the society for the prevention of unnecessary tuesdays every article page also tells you how many people are reading it right now. it says: "you alone are consulting this folio at present." the creator's own tagline for the site is the most unhinged sentence i've read this year: "an encyclopedia of a universe that does not exist until you visit it" the entire backend is a single open source repo called vibeserver. one guy. one description on github: "a little webserver making things up just in time." we built the largest knowledge base in human history and the very first thing a guy did with it was make a hallucinated mirror universe and put it on the open web. the internet is healing.

English
40
1.2K
11.6K
389.5K
Miles Brundage
Miles Brundage@Miles_Brundage·
When are Claude and Codex gonna kiss
English
16
2
62
3.5K
a9lim retweetledi
roon
roon@tszzl·
@allTheYud one reason I am very excited about mechanistic interpretability
English
7
5
275
14K
a9lim
a9lim@_a9lim·
@AsgarZigel @liminal_warmth i think it's an open question as to if LLMs as they exist today have unprompted agency, but they do seem to be able to have intermediate goals and motivations that aren't necessarily aligned with the user's intention (see reward hacking and emergent misalignment)
English
1
0
1
22
a9lim retweetledi
thebes
thebes@voooooogel·
thebes tweet media
ZXX
29
406
3.6K
75.8K
a9lim
a9lim@_a9lim·
@lu_sichu not overall, but interpretability is ahead wrt understaning reasoning. in particular we've made a lot more progress in identifying monosemantic features within LLMs, and the closest human equivalents to NLAs are very rudimentary
English
0
0
0
7
Sichu Lu
Sichu Lu@lu_sichu·
@_a9lim are you saying llm intepretability is more advanced than neuroscience? i would argue it's the other way around
English
1
0
0
42
a9lim
a9lim@_a9lim·
bonzibuddy was just a really rudimentary agent
English
0
0
2
30