can (@can) header
can (@can) profile picture

doing stuff at @modal

Joined May 2008
Photo by can (@can)Photo by can (@can)Photo by can (@can)Photo by can (@can)Photo by can (@can)Photo by can (@can)Photo by can (@can)Photo by can (@can)Photo by can (@can)Photo by can (@can)
Pinned Tweet
got a promotion today!!
1
65
8,468
can retweeted
williamsburg is over man (says the guy who works at an ai infra firm)
2
1
60
2,503
williamsburg is over man (says the guy who works at an ai infra firm)
2
1
60
2,503
I blame @paulgb
1
352
we got GTA 6 *and* Riemanns Hypotheiss before a potential Databricks IPO. At this point, it’s either human extinction or heat death of the universe that’ll come after.
Sources: Databricks is in early talks to raise several billion dollars at a ~$225B valuation, after closing a $5B round at a $190B valuation in August (Wall Street Journal) (Visit Techmeme dot com for the link and full context!)
1
2
60
2,772
guess humans were the OG stochastic parrots after all
Like most mathematicians, I believed the Riemann Hypothesis to be completely out of reach at present (even with the best AI tools), because I expected it to require substantial theory-building, probably in arithmetic geometry. The kind of new theory-building that fills entire books, and that AI isn't so good at yet. Mathematicians have a notion of how deep a problem is. A sufficiently deep problem is generally intractable until enough of the theory is in place. If difficult problems are like valuable mineral reserves buried miles underground, you first need to build the infrastructure to reach them. What Levent's proof seems to have done instead is something completely different. It takes an incredibly complicated tour through the infrastructure built for a hundred other mineral reserves, situated far from the destination, and somehow create a wormhole to the original destination. This is not theory-building in the usual sense. It is something much weirder and more alien. The best human mathematicians know how to build theory, painstakingly, to reach a goal. But no human brain, on its own, is capable of creating a wormhole that requires a 25,000 page long tour through existing infrastructure with no obvious or realistic connection to the goal. But huge credit to Levent here: I think no current AI, on its own, is capable of this either. I had never worked on or thought too much about the Riemann Hypothesis: it seemed very far removed from my own area of number theory. But the astonishing thing about Levent's claimed proof of RH (which I think has a very good chance of being correct) is that it uses a huge number of precisely the things that are central to my own corner of number theory: - Theta functions and other automorphic forms on unitary groups - Gan–Gross–Prasad period formulas - Relative trace formulas Etc., etc. A problem that I had always assumed would require new theories in arithmetic geometry may instead turn out to be tractable through an inhuman chain of existing (deep) results in higher-rank automorphic forms (and a lot else).
695
ai slop >> riemanns hypothesis
1
4
1,341
turkish mogged also lmao
1
15
582
congrats to anthropic. still remember how sonnet just felt like a breath of fresh air…you guys are good at your jobs. enjoy it.
1
1
86
2,819
it’d nice if you didn’t buy housing though
4
395
confirmed
did slack change its rendering stack recently?
5
1,184
the best essay on GTA has already been written by Tom Bissell theguardian.com/theobserver/…
43
419
56
8,460
764,030
not an original thought by a light year but still feels unclear to me when i should use astra vs sol (or opus vs fable)
Some of you have been asking very politely. Some of you have just been asking. GPT-6.2 Astra and Sol are now available. Super efficient and built to perform where it counts—the real world. openai.com/index/introducing…
8
1
35
3,106
the only movie i watched on a vision pro was tenet and i can confirm pattison is the goat
6
891
A typical sandbox spends more than half its life using less than 6% of its peak CPU. At Modal, we see that pattern for coding agents, RL rollouts, and evals. Reserve for the peak, and most of what you pay for sits idle. Modal Sandboxes burst instead, and you pay for what they use.
1
3
1
76
16,048
Replying to @paulgb
Brother…I would copy paste sudo commands I found on the internet if they hinted at fixing my python path back in the day. I really don’t think most ppl are taking the time to do due diligence.
1
8
526
Why do anything good or original when you can just use AI? 💡
9
2
69
4,321
Replying to @can
I just…dont care anymore. Let the agent figure it out
7
14
2
203
25,557
can retweeted
you know i like tuis as much as the other nerd but claude code desktop / codex.app are so good now and the sessions are actually quite rich with inline screenshots and charts and whatever — really seems we are slowly arriving at what could be a new IDE setup
24
1
214
15,119
you know i like tuis as much as the other nerd but claude code desktop / codex.app are so good now and the sessions are actually quite rich with inline screenshots and charts and whatever — really seems we are slowly arriving at what could be a new IDE setup
24
1
214
15,119
i think devin / t3 are kinda the state of the art here and bb is also very good and im sure orca / paseo etc etc are great in their own way. fast evolving, very fun and interesting!
2
6
863
some thoughts - number of agents: irrelevant, implementation detail. think in sessions. - subagents are good for long running bash / etc stuff, but less so with actual coding stuff. - local dev is obvi dead; comparatively. - irrefutable / verifiable evidence is key.
2
14
738