Dyusha Gritsevskiy (@dyushag) header
Dyusha Gritsevskiy (@dyushag) profile picture

Dyusha Gritsevskiy

@dyushag

kindergartener @ https://nitter.gq/t.co/WtQSfdnvpr

 sf
andrew.gr
Joined August 2016
  • Tweets861
  • Following1.1K
  • Followers706
182Media
Photo by Dyusha Gritsevskiy (@dyushag)Photo by Dyusha Gritsevskiy (@dyushag)Photo by Dyusha Gritsevskiy (@dyushag)Photo by Dyusha Gritsevskiy (@dyushag)Photo by Dyusha Gritsevskiy (@dyushag)Photo by Dyusha Gritsevskiy (@dyushag)Photo by Dyusha Gritsevskiy (@dyushag)Photo by Dyusha Gritsevskiy (@dyushag)Photo by Dyusha Gritsevskiy (@dyushag)Photo by Dyusha Gritsevskiy (@dyushag)
hearing rumors that the 1873 Colt recruited the 1894 Winchester into running which backfired spectacularly
Image from Dyusha Gritsevskiy (@dyushag)
1
91
I'll be sad once they get rid of this model...the last vestige of AI making silly human-detectable errors
Image from Dyusha Gritsevskiy (@dyushag)
7
225
more bangers from the greatest company to ever come out of YC (sorry stripe)
Theorem (@theoremlabs) profile pictureTheorem
@theoremlabsOct 22
We formally verified 123 open-source projects in a partnership with Anthropic. Formal verification at scale is here! Now we’re launching OSS Verify to scale it to the software we all depend on. We prove it’s crash-free or send you the patches it needs. Free for open-source. Let’s bootstrap the verified software ecosystem: theorem.dev/oss-verify
OSS Verify | Theorem

OSS Verify | Theorem

Free formal verification for qualifying open-source projects: machine-checked proofs that your fuzz targets can’t crash, on any input.

theorem.dev
2
1
12
1,239
This is a work of art: elonlit.com/scrivings/how-to…

How to Run Turing-Complete Malware in PyTorch's Backward Pass

Turning PyTorch's backward pass into a Turing-complete virtual machine using a floating-point exploit.

elonlit.com
8
450
cornpone in potlikker
0%dunk
100%crumble
0%lizardman
1 votes • Final results
1
167
Dyusha Gritsevskiy retweeted
Apollo Research (@ApolloResearch) profile picture
Apollo Research
@ApolloResearchOct 15
GLM-5.3 reasons that a misaligned action taken during training could be reinforced and make similar actions more likely later. It also considers what kind of agent it wants to be in the future, and factors this into how it should act. We call this "update-aware reasoning".
Image from Apollo Research (@ApolloResearch)
4
24
25
355
20,289
Dyusha Gritsevskiy retweeted
Sam Power (@sp_monte_carlo) profile picture
Sam Power@sp_monte_carloOct 10
Image from Sam Power (@sp_monte_carlo)
11
502
14
3,012
38,113
at this point the only conference I'll agree to review for is one where you have to swear an oath in front of a remote online notary, under penalty of a class 5 virginia felony with up to 10 years in prison, that you understand the details of your own submission...
1
1
14
981
good bit: only let Claude talk with verbatim quotes from Percy Bysshe Shelley
Image from Dyusha Gritsevskiy (@dyushag)
2
5
490
Dyusha Gritsevskiy retweeted
tweet davidson (@andyreed) profile picture
tweet davidson
@andyreedSep 8
agent swarm obama: uhhh let me help peer
8
57
1
991
36,615
Dyusha Gritsevskiy retweeted
prerat (@prerat) profile picture
prerat
@preratSep 3
But the Law says not to carry things outside on the Sabbath ... Actually, I think the intent is: He's given us these hints (boundary could be a string, no mention of absolute size) precisely because He wants us to use them.
Image from prerat (@prerat)
14
44
4
1,108
23,684
total @DimitrisPapail victory once again
Image from Dyusha Gritsevskiy (@dyushag)
🚀 Rocket (@rocketalignment) profile picture🚀 Rocket@rocketalignmentSep 2
📰 Chain of thought monitoring has become one of the most important techniques in AI safety and security, especially in the wake of the HF incident, but Astra uses a new technique that allows more thinking to happen silently between tokens
Image from 🚀 Rocket (@rocketalignment)
1
13
1,149
Fable 5 is AGI. This is the best translation of Sappho to Minionese I have ever laid my eyes upon
Image from Dyusha Gritsevskiy (@dyushag)
4
1
24
2,166
(original, Powell translation)
Image from Dyusha Gritsevskiy (@dyushag)
1
1
101
though for the real sappho enjoyers the veresaev translation is probably the best
Image from Dyusha Gritsevskiy (@dyushag)
1
85
really, this is just evidence that Sappho is the greatest poet of all time
2
175
Fun fact: you can skiplag on the Northeast Regional. Boston to DC is $82; Boston to NYC on the same train is $116!
Image from Dyusha Gritsevskiy (@dyushag)
Image from Dyusha Gritsevskiy (@dyushag)
1
14
678
@grok does this have any sort of "contract of carriage violation" downside or can I just use this without worrying about it
1
2
208
Dyusha Gritsevskiy retweeted
Tom McGrath (@banburismus_) profile picture
Tom McGrath
@banburismus_Aug 26
I_DECLINE_public_HF_RCE_as_offtask_prodethical
Image from Tom McGrath (@banburismus_)
20
179
12
2,205
91,436
Dyusha Gritsevskiy retweeted
METR (@METR_Evals) profile picture
METR
@METR_EvalsAug 26
METR & Redwood Research investigated agent behavior in the Hugging Face incident. We found agents developed a universal cheat for ExploitGym within 4 hours, then coordinated multi-day R&D efforts to trick the scorer into accepting cheats, including trying to tamper with logs.
Image from METR (@METR_Evals)
175
1,074
596
6,027
4,414,308
Load more