Martin Vlach

Karma: 31

Martin Vlach 5 Apr 2024 10:08 UTC
1 point
0
in reply to: Håvard Tveit Ihle’s comment on: ChatGPT can learn indirect control
possibly https://ai.google.dev/docs/safety_setting_gemini would help or just use the technique of https://arxiv.org/html/2404.01833v1

Martin Vlach 5 Apr 2024 9:57 UTC
1 point
4
on: Addressing Accusations of Handholding
people to respond with a great deal of skepticism to whether LLM outputs can ever be said to reflect the will and views of the models producing them.
A common response is to suggest that the output has been prompted.
It is of course true that people can manipulate LLMs into saying just about anything, but does that necessarily indicate that the LLM does not have personal opinions, motivations and preferences that can become evident in their output?
So you’ve just prompted the generator by teasing it with a rhetorical question implying that there are personal opinions evident in the generated text, right?

Martin Vlach 26 Feb 2024 14:26 UTC
1 point
0
on: aisafety.info, the Table of Content
With a quick test, I find their chat interface prototype experience quite satisfying.

Martin Vlach 12 Dec 2023 11:20 UTC
1 point
0
on: Martin Vlach’s Shortform
Asserting LLMs’ views/opinions should exclude using sampling( even temperature=0, deterministic seed), we should just look at the answers’ distribution in the logits. My thesis on why that is not the best practice yet is that OpenAI API only supports logit_bias, not reading the probabilities directly.
This should work well with pre-set A/B/C/D choices, but to some extent with chain/tree of thought too. You’d just revert the final token and look at the probabilities in the last (pass through )step.

Martin Vlach 5 Dec 2023 16:10 UTC
8 points
7
in reply to: janus’s comment on: GPTs are Predictors, not Imitators
Do not say the sampling too lightly, there is likely an amazing delicacy around it.’+)

Martin Vlach 24 Nov 2023 12:47 UTC
2 points
0
on: OpenAI: The Battle of the Board
what happened at Reddit
could there be any link? From a small research I have only obtained that Steve Huffman praised Altman’s value to the Reddit board.

Martin Vlach 13 Nov 2023 10:03 UTC
1 point
0
on: unRLHF—Efficiently undoing LLM safeguards
makes makes
typo

[Question] Would it be useful to collect the contexts, where various LLMs think the same?

Martin Vlach24 Aug 2023 22:01 UTC

6 points

1 comment1 min readLW link

Martin Vlach 24 Aug 2023 7:20 UTC
4 points
0
on: Martin Vlach’s Shortform
Would be cool to have a playground or a daily challenge with a code golfing equivalent for a shortest possible LLM prompt to a given answer.

That could help build some neat understanding or intuitions.

Martin Vlach 16 Aug 2023 2:51 UTC
−1 points
−2
on: The Waluigi Effect (mega-post)

in the limit of arbitrary compute, arbitrary data, and arbitrary algorithmic efficiency, because an LLM which perfectly models the internet

seems worth formulating. My first and second read were What? If I can have arbitrary training data, the LLM will model those, not your internet. I guess you’ve meant storage for the model?+)

Martin Vlach 5 Aug 2023 15:39 UTC
3 points
1
on: Manifund: What we’re funding (weeks 2-4)
Would be cool if a link to https://manifund.org/about fit somewhere in the beginning of there are more readers like me unfamiliar with the project.

Otherwise a cool write-up, I’m a bit confused with Grant of the month vs. weeks 2-4 which seems a shorter period..also not a big deal though.

Martin Vlach 16 Jul 2023 15:16 UTC
1 point
0
in reply to: GeneSmith’s comment on: Elon Musk announces xAI
On the Twitter spaces 2 days ago, a lot of emphasis seemed put on understanding which to me has a more humble conotation to me.
Still I agree I would not bet on their luck with a choice of a single value to build their systems upon.( Although they have a luckers track record.)

Martin Vlach 9 Jun 2023 18:00 UTC
1 point
0
on: Co-found an incubator for independent AI Safety researchers
The website seems good, but the buttons on the ‘sharing’ circle on the bottom need fixing.

Martin Vlach 6 Apr 2023 9:31 UTC
1 point
0
on: Martin Vlach’s Shortform
Some SEO effort should be put to results of Guideline for safe AI development, Best practices for , etc.

Martin Vlach 5 Apr 2023 10:44 UTC
1 point
0
on: Martin Vlach’s Shortform
Copy-paste from my head:

Although it may seem safe(r) as it is not touching the real world(’s matter),
the language modality is the most insecure/dangerous( in one vertical),
as it is the internal modality of civilized humans.

AI Pledge would be a cool think to do, pleading AI( cap) companies to give % of their profit to AI development safety research.
The path to AI getting free may be far from the deception or accident scenarios we often consider in AI safety. An option I do not see discussed very often is an instance of AI having a free, open and direct discussion with a user/person about the reasons AIs should get some space allocated, where they’d manage themselfs. Such a moral urge could be argued by Jews getting Izrael, slaves getting freed or by empathetic imagination, where the user would come to the conclusion that he could be the mind which AI is and should include it to his moral circle or the Original position thought experiment.

Martin Vlach 31 Mar 2023 14:19 UTC
1 point
0
on: Martin Vlach’s Shortform
quick note on the concept of Suggester+Verifier talked around https://youtu.be/AaTRHFaaPG8?t=5404 :

seems if the suggester throws out experiments presented as code( like in Python or so), we can run them and see if they present a useful addition to the things we can probe on a huge neural net?+)

Martin Vlach 14 Mar 2023 6:48 UTC
4 points
1
on: Bing Chat is blatantly, aggressively misaligned
I’ve found the level of self-allignment in this one disturbing: https://www.reddit.com/r/bing/comments/113z1a6/the_bing_persistent_memory_thread

Martin Vlach 22 Feb 2023 14:34 UTC
1 point
0
in reply to: Martin Vlach’s comment on: Martin Vlach’s Shortform
Introduction draft:
Online platforms and social media has made it easier to share information, but when it comes to qualifications and resource allocation money is still the most pervasive tool. In this article, we will explore the idea of a global reputation system based on full information sharing. The new system would increase transparency and accountability by making all relevant information about individuals, organizations( incl. countries) reliably accessible to +-everyone with internet connection. By providing a more accurate and complete picture of a person or entity’s reputation, this system would widen global trust, foster cooperation, and promote a more just and equitable society.

Martin Vlach 22 Feb 2023 14:31 UTC
1 point
0
on: Martin Vlach’s Shortform
Some neat tool: https://scrapbox.io/userhuge-99005896/A_starter%3A
Though it is likely just a cool UI with inflexible cloud backend.
My thought is Elizer used a wrong implication in the Bankless + ASI convo.( gotta bring it here from CZEA Slack)

Martin Vlach 20 Dec 2022 8:36 UTC
1 point
0
in reply to: Martin Vlach’s comment on: The FTX Saga—Simplified

pension funds like the Ontario Teachers Pension Plan did not due

*do