TurnTrout

Karma: 19,839

My name is Alex Turner. I’m a research scientist at Google DeepMind on the Scalable Alignment team. My views are strictly my own; I do not represent Google. Reach me at alex[at]turntrout.com

Steering Gemini with BiDPO

TurnTrout31 Jan 2025 2:37 UTC

83 points

3 comments1 min readLW link

(turntrout.com)

Insights from “The Manga Guide to Physiology”

TurnTrout24 Jan 2025 5:18 UTC

26 points

3 comments1 min readLW link

(turntrout.com)

Deceptive Alignment and Homuncularity

Oliver Sourbut and TurnTrout

16 Jan 2025 13:55 UTC

25 points

12 comments22 min readLW link

Gaming TruthfulQA: Simple Heuristics Exposed Dataset Weaknesses

TurnTrout16 Jan 2025 2:14 UTC

64 points

3 comments1 min readLW link

(turntrout.com)

Review: Breaking Free with Dr. Stone

TurnTrout18 Dec 2024 1:26 UTC

47 points

5 comments1 min readLW link

(turntrout.com)

Gradient Routing: Masking Gradients to Localize Computation in Neural Networks

cloud, Jacob G-W, Evzen, Joseph Miller and TurnTrout

6 Dec 2024 22:19 UTC

161 points

12 comments11 min readLW link

(arxiv.org)

Deep Causal Transcoding: A Framework for Mechanistically Eliciting Latent Behaviors in Language Models

Andrew Mack and TurnTrout

3 Dec 2024 21:19 UTC

95 points

7 comments41 min readLW link

Intrinsic Power-Seeking: AI Might Seek Power for Power’s Sake

TurnTrout19 Nov 2024 18:36 UTC

40 points

5 comments1 min readLW link

(turntrout.com)

Announcing turntrout.com, my new digital home

TurnTrout17 Nov 2024 17:42 UTC

107 points

24 comments1 min readLW link

(turntrout.com)

I found >800 orthogonal “write code” steering vectors

Jacob G-W and TurnTrout

15 Jul 2024 19:06 UTC

99 points

19 comments7 min readLW link

(jacobgw.com)

Mechanistically Eliciting Latent Behaviors in Language Models

Andrew Mack and TurnTrout

30 Apr 2024 18:51 UTC

206 points

43 comments45 min readLW link

Many arguments for AI x-risk are wrong

TurnTrout5 Mar 2024 2:31 UTC

158 points

87 comments12 min readLW link

Dreams of AI alignment: The danger of suggestive names

TurnTrout10 Feb 2024 1:22 UTC

103 points

59 comments4 min readLW link

Steering Llama-2 with contrastive activation additions

Nina Panickssery, Wuschel Schulz, NickGabs, Meg, evhub and TurnTrout

2 Jan 2024 0:47 UTC

124 points

29 comments8 min readLW link

(arxiv.org)

How should TurnTrout handle his DeepMind equity situation?

habryka and TurnTrout

16 Oct 2023 18:25 UTC

63 points

35 comments6 min readLW link 1 review

Paper: Understanding and Controlling a Maze-Solving Policy Network

TurnTrout, Ulisse Mini, peligrietzer, mrinank_sharma, Austin Meek, Monte M and lisathiergart

13 Oct 2023 1:38 UTC

70 points

0 comments1 min readLW link

(arxiv.org)

AI presidents discuss AI alignment agendas

TurnTrout and Garrett Baker

9 Sep 2023 18:55 UTC

217 points

23 comments1 min readLW link

(www.youtube.com)

ActAdd: Steering Language Models without Optimization

technicalities, TurnTrout, lisathiergart, David Udell, Ulisse Mini and Monte M

6 Sep 2023 17:21 UTC

105 points

3 comments2 min readLW link

(arxiv.org)

Open problems in activation engineering

TurnTrout, woog, lisathiergart, Monte M and Ulisse Mini

24 Jul 2023 19:46 UTC

51 points

2 comments1 min readLW link

(coda.io)

Ban development of unpredictable powerful models?

TurnTrout20 Jun 2023 1:43 UTC

46 points

25 comments4 min readLW link