eigenron

metatheorist

[eigenron @eigenron]

Senior Year Undergrad Research on Phase Qubits FINALLY DONE.

> derived hamiltonian for a Josephson junction-based phase qubit
> mapped it to a spin-1/2 system hamiltonian under magnetic fields
> studied the quantum dynamics and evolution of the hamiltonian
> derived spin-flip probabilities
> explored qubit control via phase shift and applied mag fields for high fidelity

[eigenron @eigenron]

wake up babe all of twitter is now bullish on sims

[mgostIH @mgostIH]

P = NP is true because you can reduce any problem to matrix multiplication

[David @DavidSHolz]

@Karmedge no one ever seems to ask "what if we put a billion dollars into simulators instead of data collection"

[David Deutsch @DavidDeutschOxf]

Constructor theory of time, published version: https://t.co/JG3oLvyNuY

[eigenron @eigenron]

love

[eigenron @eigenron]

one of the more interesting thing's i've read this year

[Dwarkesh Patel @dwarkesh_sp]

This got totally buried cause it came out the same day as Navier Stokes.

But I think this is super interesting evidence about what the true major source of algorithmic progress is.

[eigenron @eigenron]

@reactorworld @nvidia awesome!! i can't seem to access the repo though - can you guys make it public?

[Curt Tigges @CurtTigges]

We recently went viral on X with the claim that neurons are "already clean." But what do we mean by that? Are neurons are a useful basis for interp? We're finally releasing our technical report on LLM neurons--and what we've found is far more interesting than mono/polysemanticity.https://t.co/4ELaLkYMrf

[eigenron @eigenron]

has anyone here used SoftMimicGen for sim data curation? would love to chat! DMs open

[eigenron @eigenron]

@yacinelearning https://t.co/EB7u0XVEHh

[eigenron @eigenron]

@tanayj too slow and insanely bloated in reasoning tokens - i tried on max effort following their blog, albeit wihout any harness. probably will try again with our harnesses to test again. but (action) inference is too too slow.

[eigenron @eigenron]

i tried it on a box folding task and i was … disappointed (to say the least)

[Claude @claudeai]

Fable 5.2 shows substantially improved skills in controlling physical robots, even on tasks it has never explicitly been trained to do.

On RoboDojo, a benchmark with a simulated two-arm robot, it outperformed other models and most AI systems built to control robots.

[eigenron @eigenron]

@dariusfdi very much so

[eigenron @eigenron]

opus5.5 is the best model i've ever used. for the first time i feel like i'm getting useful work done while also understanding and building things on my own (post-claude code). i love it.

[eigenron @eigenron]

@tanayj too slow and insanely bloated in reasoning tokens - i tried on max effort following their blog, albeit wihout any harness. probably will try again with our harnesses to test again. but (action) inference is too too slow.

[eigenron @eigenron]

did they already nerf opus5.5?

[eigenron @eigenron]

wake up babe new robotic model just dropped

[Claude @claudeai]

Fable 5.2 shows substantially improved skills in controlling physical robots, even on tasks it has never explicitly been trained to do.

On RoboDojo, a benchmark with a simulated two-arm robot, it outperformed other models and most AI systems built to control robots.

[keerti manney✈️CoRL’26 @keertimanney]

I'll be in Austin for CoRL this week. Reach out if you want to talk simulations, agentic robotics, and what it takes to deploy robots to work outside the lab. DMs open!

[Shrey Kothari @CoRL @Shreyko]

I'll be in Austin for CoRL. DM me if you want to chat simulations, building or deploying robots, or know any good BBQ spots!

[Boaz Barak @boazbaraktcs]

An update on my AI safety course with links to some videos and scribe notes.

https://t.co/jNo1rhuBHD

[eigenron @eigenron]

maximally truth seeking = indifferent to human existence