Sam Marks

Karma: 3,185

Modifying LLM Beliefs with Synthetic Document Finetuning

RowanWang, Johannes Treutlein, Avery, Ethan Perez, Fabien Roger and Sam Marks

Apr 24, 2025, 9:15 PM

69 points

12 comments2 min readLW link

(alignment.anthropic.com)

Downstream applications as validation of interpretability progress

Sam MarksMar 31, 2025, 1:35 AM

112 points

3 comments7 min readLW link

Auditing language models for hidden objectives

Sam Marks, Johannes Treutlein, dmz, Sam Bowman, Hoagy, Carson Denison, Kei, 7vik, Akbir Khan, Austin Meek, Euan Ong, Christopher Olah, Fabien Roger, jeanne_, Meg, Drake Thomas, Adam Jermyn, Monte M and evhub

Mar 13, 2025, 7:18 PM

141 points

15 comments13 min readLW link

Recommendations for Technical AI Safety Research Directions

Sam MarksJan 10, 2025, 7:34 PM

64 points

1 comment17 min readLW link

(alignment.anthropic.com)

Alignment Faking in Large Language Models

ryan_greenblatt, evhub, Carson Denison, Benjamin Wright, Fabien Roger, Monte M, Sam Marks, Johannes Treutlein, Sam Bowman and Buck

Dec 18, 2024, 5:19 PM

483 points

75 comments10 min readLW link

SAEBench: A Comprehensive Benchmark for Sparse Autoencoders

Can, Adam Karvonen, Johnny Lin, Curt Tigges, Joseph Bloom, chanind, Yeu-Tong Lau, Eoin Farrell, Arthur Conmy, CallumMcDougall, Kola Ayonrinde, Matthew Wearden, Sam Marks and Neel Nanda

Dec 11, 2024, 6:30 AM

82 points

6 comments2 min readLW link

(www.neuronpedia.org)

Evaluating Sparse Autoencoders with Board Game Models

Adam Karvonen, Sam Marks, Can, Benjamin Wright, Jannik Brinkmann, Logan Riggs and Rico Angell

Aug 2, 2024, 7:50 PM

38 points

1 comment9 min readLW link

Discriminating Behaviorally Identical Classifiers: a model problem for applying interpretability to scalable oversight

Sam MarksApr 18, 2024, 4:17 PM

113 points

10 comments12 min readLW link

What’s up with LLMs representing XORs of arbitrary features?

Sam MarksJan 3, 2024, 7:44 PM

158 points

63 comments16 min readLW link

Some open-source dictionaries and dictionary learning infrastructure

Sam MarksDec 5, 2023, 6:05 AM

46 points

7 comments5 min readLW link

Thoughts on open source AI

Sam MarksNov 3, 2023, 3:35 PM

62 points

17 comments10 min readLW link

Turning off lights with model editing

Sam Marks12 May 2023 20:25 UTC

68 points

5 comments2 min readLW link

(arxiv.org)

[Crosspost] ACX 2022 Prediction Contest Results

Scott Alexander, Eric Neyman and Sam Marks

24 Jan 2023 6:56 UTC

48 points

6 comments8 min readLW link

AGISF adaptation for in-person groups

Sam Marks, Xander Davies and Richard_Ngo

13 Jan 2023 3:24 UTC

44 points

2 comments3 min readLW link

Update on Harvard AI Safety Team and MIT AI Alignment

Xander Davies, Sam Marks, kaivu, tlevin, eleni, maxnadeau and Naomi Bashkansky

2 Dec 2022 0:56 UTC

60 points

4 comments8 min readLW link

Recommend HAIST resources for assessing the value of RLHF-related alignment research

Sam Marks and Xander Davies

5 Nov 2022 20:58 UTC

26 points

9 comments3 min readLW link

Caution when interpreting Deepmind’s In-context RL paper

Sam Marks1 Nov 2022 2:42 UTC

105 points

8 comments4 min readLW link

Safety considerations for online generative modeling

Sam Marks7 Jul 2022 18:31 UTC

42 points

9 comments14 min readLW link

Proxy misspecification and the capabilities vs. value learning race

Sam Marks16 May 2022 18:58 UTC

23 points

3 comments4 min readLW link

If you’re very optimistic about ELK then you should be optimistic about outer alignment

Sam Marks27 Apr 2022 19:30 UTC

17 points

8 comments3 min readLW link