Linda Linsefors

Featured in:

Articles

Is the output of the softmax in a single transformer attention head usually winner-takes-all? — LessWrong

2 months ago | lesswrong.com | Linda Linsefors

Using the notation from here: A Mathematical Framework for Transformer CircuitsThe attention pattern for a single attention head is determined by A=softmax(xTWTQWKx), where softmax is computed for each row of xTWTQWKx.Each row of A gives the attention pattern for the current token. Are these rows (post softmax) typically close to one-hot? I.e. are they mainly dominated by a single attention (per current token).
Theory of Change for AI Safety Camp — LessWrong

Jan 22, 2025 | lesswrong.com | Linda Linsefors

In a private discussion related to our fundraiser it was pointed out to me that AISC has not made clear enough what our actual theory of change is. Therefore this post. Some caveats/context:This is my personal viewpoint. Other organisers might disagree about what is central or not. I’ve co-organised AISC1, AISC8, AISC9, and now AISC10. Remmelt has co-organised all except AISC2. Robert recently joined for AISC10. I hope there will be an AISC11, but in either case, I will no longer be an organiser.
We don't want to post again "This might be the last AI Safety Camp" — LessWrong

Jan 21, 2025 | lesswrong.com | Linda Linsefors |Robert Kralisch

We still need more funding to be able to run another edition. Our fundraiser raised $6k as of now, and will end if it doesn't reach the $15k minimum, on February 1st. We need proactive donors. If we don't get funded for this time, there is a good chance we will move on to different work in AI Safety and new commitments. This would make it much harder to reassemble the team to run future AISCs, even if the funding situation improves.
AI Safety Camp 10 — LessWrong

Oct 26, 2024 | lesswrong.com | Robert Kralisch |Linda Linsefors

We are pleased to announce that the 10th version of the AI Safety Camp is now entering the team member application phase!We again have a wide range of projects this year, so check them out to see if you or someone you know might be interested in applying to join one of them. You can find all of the projects and the application form on our website, or directly apply here. The deadline for team member applications is November 17th (Sunday).
Inner Alignment - LessWrong

Apr 1, 2024 | lesswrong.com | Linda Linsefors |Seth Herd |Rob Bensinger
Some costs of superposition — LessWrong

Mar 3, 2024 | lesswrong.com | Linda Linsefors |Charlie Steiner

I don't expect this post to contain anything novel. But from talking to others it seems like some of what I have to say in this post is not widely known, so it seemed worth writing. In this post I'm defining superposition as: A representation with more features than neurons, achieved by encoding the features as almost orthogonal vectors in neuron space. One reason to expect superposition in neural nets (NNs), is that for large n, Rn has many more than n almost orthogonal directions.
This might be the last AI Safety Camp — LessWrong

Jan 24, 2024 | lesswrong.com | Linda Linsefors |Thomas Kwa |Lucius Bushnaq |Simon Fischer

MATS has steadily increased in quality over the past two years, and is now more prestigious than AISC. We also have Astra, and people who go directly to residencies at OpenAI, Anthropic, etc. One should expect that AISC doesn't attract the best talent. There is so much wrong here, I don't even know how to start (i.e. I don't know what the core cruxes are) but I'll give it a try. I AISC is not MATS because we're not trying to be MATS.
Rates of Criminality Amongst Giving Pledge Signatories — EA Forum

Jan 22, 2024 | forum.effectivealtruism.org | Ulrik Horn |Robi Rahman |Nathan Young |Linda Linsefors

The base rate seems tricky to interpret without a real comparison. 3 out of 5 points in the summary say that the rate is high, but it doesn't seem like the evidence exists to support that claim. The implied reference class is "non EA super-rich people", but that class isn't actually examined in the data. The rates do seem subjectively very high.
Interpreting the Learning of Deceit — LessWrong

Dec 18, 2023 | lesswrong.com | Roger Dearnaley |Linda Linsefors |Bogdan Ionut Cirstea |Neel Nanda

One of the primary concerns when attempting to control an AI of human-or-greater capabilities is that it might be deceitful. It is, after all, fairly difficult for an AI to succeed in a coup against humanity if the humans can simply regularly ask it "Are you plotting a coup? If so, how can we stop it?" and be confident that it will give them non-deceitful answers! TL;DR LLMs demonstrably learn deceit from humans.