RSS

Su­per­po­si­tion is not “just” neu­ron polysemanticity

LawrenceC26 Apr 2024 23:22 UTC
32 points
0 comments13 min readLW link

D&D.Sci Long War: Defen­der of Data-mocracy

aphyer26 Apr 2024 22:30 UTC
28 points
0 comments3 min readLW link

On Not Pul­ling The Lad­der Up Be­hind You

Screwtape26 Apr 2024 21:58 UTC
52 points
3 comments9 min readLW link

We are headed into an ex­treme com­pute overhang

devrandom26 Apr 2024 21:38 UTC
20 points
5 comments2 min readLW link

[Con­cept Depen­dency] Edge Reg­u­lar Lat­tice Graph

Johannes C. Mayer26 Apr 2024 21:14 UTC
5 points
0 comments1 min readLW link

[Con­cept Depen­dency] Con­cept Depen­dency Posts

Johannes C. Mayer26 Apr 2024 20:57 UTC
8 points
2 comments2 min readLW link

[Question] Wouldn’t weak AI agents provide warn­ing?

Mandatory Topic26 Apr 2024 19:34 UTC
5 points
0 comments1 min readLW link

Duct Tape security

Isaac King26 Apr 2024 18:57 UTC
60 points
7 comments5 min readLW link

Fun­da­men­tal Uncer­tainty: Chap­ter 8 - When does fun­da­men­tal un­cer­tainty mat­ter?

Gordon Seidoh Worley26 Apr 2024 18:10 UTC
9 points
2 comments32 min readLW link

Scal­ing of AI train­ing runs will slow down af­ter GPT-5

Maxime Riché26 Apr 2024 16:05 UTC
32 points
5 comments3 min readLW link

Spa­tial at­ten­tion as a “tell” for em­pa­thetic simu­la­tion?

Steven Byrnes26 Apr 2024 15:10 UTC
50 points
7 comments8 min readLW link

Arch-anarchy

Peter lawless 26 Apr 2024 15:05 UTC
−1 points
1 comment25 min readLW link

An In­tro­duc­tion to AI Sandbagging

Teun van der Weij, Felix Hofstätter and Francis Rhys Ward
26 Apr 2024 13:40 UTC
28 points
0 comments8 min readLW link

LLMs seem (rel­a­tively) safe

JustisMills25 Apr 2024 22:13 UTC
41 points
12 comments7 min readLW link
(justismills.substack.com)

Los­ing Faith In Con­trar­i­anism

omnizoid25 Apr 2024 20:53 UTC
38 points
25 comments5 min readLW link

Why I stopped be­ing into basin broadness

tailcalled25 Apr 2024 20:47 UTC
14 points
1 comment2 min readLW link

AXRP Epi­sode 29 - Science of Deep Learn­ing with Vikrant Varma

DanielFilan25 Apr 2024 19:10 UTC
18 points
1 comment63 min readLW link

Im­prov­ing Dic­tionary Learn­ing with Gated Sparse Autoencoders

Neel Nanda, Senthooran Rajamanoharan, Arthur Conmy, lsgos, Tom Lieberum, Vikrant Varma, János Kramár and Rohin Shah
25 Apr 2024 18:43 UTC
60 points
23 comments1 min readLW link
(arxiv.org)

“Why I Write” by Ge­orge Or­well (1946)

Arjun Panickssery25 Apr 2024 16:02 UTC
52 points
3 comments9 min readLW link
(www.orwellfoundation.com)

Knowl­edge Base 8: The truth as an at­trac­tor in the in­for­ma­tion space

iwis25 Apr 2024 15:28 UTC
−10 points
0 comments2 min readLW link