RSS

Marius Hobbhahn

Karma: 2,890

I recently founded Apollo Research: https://​​www.apolloresearch.ai/​​

I was previously doing a Ph.D. in ML at the International Max-Planck research school in Tübingen, worked part-time with Epoch and did independent AI safety research.

For more see https://​​www.mariushobbhahn.com/​​aboutme/​​

I subscribe to Crocker’s Rules

How to Sleep Better

Marius Hobbhahn16 Jul 2021 0:00 UTC
48 points
49 comments15 min readLW link

A Guide for Productivity

Marius Hobbhahn23 Jul 2021 7:03 UTC
33 points
7 comments41 min readLW link

What makes us happy and de­pressed?

Marius Hobbhahn3 Oct 2021 6:25 UTC
16 points
6 comments19 min readLW link

What are red flags for Neu­ral Net­work suffer­ing?

Marius Hobbhahn8 Nov 2021 12:51 UTC
29 points
15 comments12 min readLW link

How to mea­sure FLOP/​s for Neu­ral Net­works em­piri­cally?

Marius Hobbhahn29 Nov 2021 15:18 UTC
16 points
5 comments7 min readLW link

What’s the back­ward-for­ward FLOP ra­tio for Neu­ral Net­works?

13 Dec 2021 8:54 UTC
19 points
12 comments10 min readLW link

Causal­ity, Trans­for­ma­tive AI and al­ign­ment—part I

Marius Hobbhahn27 Jan 2022 16:18 UTC
14 points
11 comments8 min readLW link

Nu­clear En­ergy—Good but not the silver bul­let we were hop­ing for

Marius Hobbhahn30 Apr 2022 15:41 UTC
64 points
33 comments15 min readLW link1 review

The limits of AI safety via debate

Marius Hobbhahn10 May 2022 13:33 UTC
29 points
7 comments10 min readLW link

Elic­it­ing La­tent Knowl­edge (ELK) - Distil­la­tion/​Summary

Marius Hobbhahn8 Jun 2022 13:18 UTC
69 points
2 comments21 min readLW link

In­ves­ti­gat­ing causal un­der­stand­ing in LLMs

14 Jun 2022 13:57 UTC
28 points
6 comments13 min readLW link

Our men­tal build­ing blocks are more differ­ent than I thought

Marius Hobbhahn15 Jun 2022 11:07 UTC
44 points
11 comments14 min readLW link

Reflec­tion Mechanisms as an Align­ment tar­get: A survey

22 Jun 2022 15:05 UTC
32 points
1 comment14 min readLW link

What suc­cess looks like

28 Jun 2022 14:38 UTC
19 points
4 comments1 min readLW link
(forum.effectivealtruism.org)

Trends in GPU price-performance

1 Jul 2022 15:51 UTC
85 points
12 comments1 min readLW link1 review
(epochai.org)

The Defen­der’s Ad­van­tage of Interpretability

Marius Hobbhahn14 Sep 2022 14:05 UTC
41 points
4 comments6 min readLW link

Why de­cep­tive al­ign­ment mat­ters for AGI safety

Marius Hobbhahn15 Sep 2022 13:38 UTC
57 points
13 comments13 min readLW link

Paper+Sum­mary: OMNIGROK: GROKKING BEYOND ALGORITHMIC DATA

Marius Hobbhahn4 Oct 2022 7:22 UTC
46 points
11 comments1 min readLW link
(arxiv.org)

Reflec­tion Mechanisms as an Align­ment tar­get: A fol­low-up survey

5 Oct 2022 14:03 UTC
15 points
2 comments7 min readLW link

Les­sons learned from talk­ing to >100 aca­demics about AI safety

Marius Hobbhahn10 Oct 2022 13:16 UTC
214 points
17 comments12 min readLW link1 review