Vik

ML and neuroscience researcher, gamer rat ex-coach | Jungian ML essays at | projects:

[Full ML Alchemist @unironictechbro]

well

[Vik @viktor_thoth]

@daniel_mac8 I'm glad people are still reading Kahneman's book. interleaved prompt optimization on top of RL tho has nothing to do in any kind of sensible framing with System 1 and 2. citing neuro papers on top is cheap intellectual dishonesty.

[Subbarao Kambhampati (కంభంపాటి సుబ్బారావు) @rao2z]

Mechanistic Interpretability be like..

We wanted to see what the bread slice is thinking when it is in the toaster. Is it taking its singing equanimously? thinking religious thoughts? or plotting revenge? (Face it--a vengeful singed slice is the last thing we want!)

But alas, the bread slice doesn't talk! Its thoughts are intricately encoded in its singe pattern.

So we devised a clever method--of showing the bread slice to a random Joe, asking him to describe its feelings.

Joe says that the slice is having a religious experience involving Mother Mary.

But is random Joe describing these feelings right?

To check, we called random artist Jane and asked her to render these thoughts back to the external appearance of a (different!) bread slice.

We co-trained Joe and Jane until they learn to auto-encode the real truth about the slice's thoughts.

We are finding this technique to be a great way to understand what the bread slice is thinking. The tehcnique is not always (or even sometimes) correct, but it gives us a great window into publishing bread slice thought related podcast articles.

Hope you find this research useful. #AIAphorisms

[Yann LeCun @ylecun]

@nxthompson So much BS

[Matangi, the Mystic @BlumenMit]

A chemist reviewed 100 years of chemical analysis and found no evidence where the music lives in the radio

[Pascal-Emmanuel Gobry @pegobry_en]

The death of RTS as a genre is proof positive that zoomers are dumber than millennials

[Michael Andregg @michaelandregg]

We've uploaded a fruit fly. We took the @FlyWireNews connectome of the fruit fly brain, applied a simple neuron model (@Philip_Shiu Nature 2024) and used it to control a MuJoCo physics-simulated body, closing the loop from neural activation to action.

A few things I want to say about what this means and where we're going at @eonsys. 🧵

[calle @callebtc]

I'm fascinated by this level of existential crisis developers seem to be going through.

The uncomfortable truth is nobody needs you to be an artisan coder.

Nobody cares about how you coded your app, or whether you feel an emotional attachment to your craft.

You were always code monkey with a high enough salary to believe that your individualist craftsmanship matters to anyone.

It doesn't matter to anyone but you. Not your employer, not your customer. Nobody cares about how you made the product. Nobody cares about your attachment to your process.

You're experiencing the same as countless other artisans have experienced in the last century.

I'm happy for you. You were starting to believe that you're a demigod amongst mortals.

You're not. A machine is better than you.

Now you're free.

[⭕ Brock Pierson @brockpierson]

Openclaw is one big grift.

Nobody is building anything real.

It's just grifting influencers telling YOU how to build stuff (but never build anything themselves).

Interesting how you can be building 24/7 but have nothing to show for it...

Check the track record of these snake oil salesmen.

Same type of person who polluted crypto with their grifty ways.

Selling a false dream and taking money from innocent people is disgusting.

You can't change my mind.

[Dr. Alex Wissner-Gross @alexwg]

https://t.co/N2AwpI9f8m

[Vik @viktor_thoth]

@zackdfilms1 rn it functions as a beer opener

[Vik @viktor_thoth]

@zackdfilms1 it's for sale, dm me and take it

[Yacine Mahdid @yacinelearning]

“if each synapses maps to a neural net weight”

[PS @dostoevesque]

There is something slightly irritating about how youngsters today seem obsessed with diagnosing an entire generation of their parents using therapy language they picked up from reels and threads. We are effectively criticizing them for failing a test they didn't even know they were taking, using vocabulary that didn't exist in their world.

It is easy to call your own father a narcissist when you have grown up listening to podcasts that explain attachment theory and emotional patterns. It is harder to imagine what it meant to be a twenty five year old parent in the nineteen eighties with no language for feelings, no examples to follow, no online spaces to learn from, no therapy culture, only responsibility and pressure.

Instead of extending the empathy we have learned through this new vocabulary to those who never had access to it, we often turn that language into a weapon against them. Social media encourages this because anger feels powerful and validating. A label like narcissist or toxic neatly explains the pain and gives it a clear target.

By framing their behavior through complex diagnostic labels they cannot define, you strip them of the ability to defend themselves or explain their perspective. They retreat into silence because they literally do not have the words to counter your argument. It allows you to dominate the interaction not because you are morally right, but because you possess a vocabulary that makes them feel stupid and inferior.

[Vik @viktor_thoth]

@r0ck3t23 a full-on autist's understanding of what language is

[Vik @viktor_thoth]

@bryan_johnson hey Bryan, could you add Hungary to the countries you deliver Blueprint to? would love to order

[Vik @viktor_thoth]

@ns123abc yeah and it will slop into degeneracy

[Vik @viktor_thoth]

@Hangsiin @GoogleDeepMind I'm surprised such an inconsequential paper with unsupported ambitious claims can come out of deepmind. they found a 1% better activation function with an expensive LLM-based search that can hardly scale to combinatorial NAS problems. this could have been a Medium post.

[Vik @viktor_thoth]

https://t.co/eQVnfqTkaQ