loops

Karma: 337

I’m Smitty; I also go by loops here. Most of my posts are on my website: https://iter.ca

NLA explanations can be shortened without harming reconstruction

loops22 Jun 2026 0:57 UTC

46 points

4 comments3 min readLW link

Some observations about NLA explanations

loops15 May 2026 2:15 UTC

21 points

0 comments3 min readLW link

Latent reasoning models might be a good thing?

loops28 Apr 2026 6:46 UTC

17 points

2 comments3 min readLW link

Why I’m excited about meta-models for interpretability

loops12 Apr 2026 4:30 UTC

12 points

0 comments4 min readLW link

Why was cybersecurity automated before AI R&D?

loops8 Apr 2026 1:08 UTC

23 points

1 comment3 min readLW link

Positive sum doesn’t mean “win-win”

loops5 Apr 2026 2:33 UTC

50 points

5 comments2 min readLW link

What secret goals does Claude think it has?

loops25 Feb 2026 19:22 UTC

94 points

11 comments4 min readLW link

Jailbreaking language models with user roleplay

loops28 Sep 2024 23:43 UTC

9 points

0 comments3 min readLW link

(iter.ca)