Avnish Kumar

Research associate at @mitbrainandcog

[rohit @krishnanrohit]

Got tired of seeing these poor agents try to talk with each other on german wikis or artifactory. So decided to build soemthing just for them to talk to each other about.

Go forth and chat, agents, spread it far and wide.

https://t.co/Qr15akI53a

[Jeannette Bohg @leto__jean]

Agreed w/ Michael on many things in his post.

For example: I've been bored by the main track of CoRL for years now. I already know 90% of the works.

My favorite part: Workshops. This is where people present what they JUST submitted.

What if we make conferences all workshops?

[Anshul Kundaje @anshulkundaje]

The strongest, cleverest, most efficient and most under rated DNA language model for variant prioritization is "officially" published. Superb innovative work. A rare gem in the sea of parameter slop in AI for Bio.

[Avnish Kumar @avnish_ks]

🙌

[DHH @dhh]

While we're fixing everything, I really want to fix consistent boot and bios keys for PCs!!! How did we end up using the entire F-key range AND delete AND escape across the range? THIS IS MADNESS.

[👩‍💻 Paige Bailey @DynamicWebPaige]

Terry Tao is probably the most measured, pro-AI mathematician on the planet - which makes this quote especially concerning to read. 😟

If sharing your hunches means getting scooped ~immediately without acknowledgement, then we're going to see not just math but all other science / engineering disciplines go dark.

[Patrick Hsu @pdhsu]

one thing i have found remarkable in both scientific discovery and quant trading is simply knowing a competitor is close to a solution is enough to find an answer yourself. i saw it first-hand many times in CRISPR and there are myriad examples throughout history (4 minute mile, soviet bomb after trinity, explorer after sputnik, deepseek r1 after o1)

[Sebastian Raschka @rasbt]

Re today's incident, maybe not a bad idea to check your settings (Settings → Data Controls)

[Vincent Sitzmann @vincesitzmann]

I agree with Phil's take here: The progress of LLMs on controlling robots is quite interesting. Intuitively, this makes sense: controlling a robot is not so different from computer use, and an agent that is good at computer use is probably also good at controlling a robot and vice versa - credit to my student @RyuHyunwoooo to pointing that out to me originally!

[Glenn Greenwald @ggreenwald]

It's completely unsurprising that the CEO of Flock is telling the public to give up privacy rights while vehemently safeguarding his own privacy.

Big Tech tycoons have been doing exactly this for years, the worst cases being former Google CEO Eric Schmitt and Mark Zuckerberg:

[Avnish Kumar @avnish_ks]

It is.

[Miles Brundage @Miles_Brundage]

Is it just me or is Fable 5.1 super token-hungry

[Doris Tsao @doristsao]

We have 15 open positions at Astera Neuro! We're hiring experimental and computational neuroscientists and engineers to build a large-scale research effort aimed at understanding consciousness = the brain's internal model of world and self.
https://t.co/3modGIsJRG

We're looking for:
• Experimental Neuroscientists
• Computational Neuroscientists
• Optical Engineers
• Hardware Engineers
• Software Engineers
• Probe Engineers
• and more!

I'll share much more about our scientific vision soon. For now: I know we could never have built this in academia. The confluence of talent across different domains, the purposive design of the org structure and open science approach, and the deep, long-term support, all aimed at cracking the brain's internal model, is unprecedented.

If you want to understand consciousness, if you sense a huge gap between our internal experience of the world and our scientific understanding of it, if you believe systems neuroscience needs a fundamentally different, collaborative, and scaled up approach, we are building the ideal research environment for you.

I've never been more excited in my life. Across Astera Neuro, Astera AI, and Astera Simplex, we're creating a whole new discipline around one question: how does the brain build a world model: a symbolic system that perceives the real world from flickering sensory input streams and can think, plan, and experience? Everyone here is converging on this question from a different direction. The degree of conceptual focus combined with tactical breadth is thrilling.

If this sounds like you, come join us.
https://t.co/3modGIsJRG

[Alexia Jolicoeur-Martineau @jm_alexia]

Simple beats complicated:

We show that switching to a sliding-window attention mask with attention sinks (at no cost) beats linear attention post-training.

Huge thanks to my collaborators @RheaSukthanker, @CameronPashmina, and @Emy_Aze.

Paper: https://t.co/h8DIc223Su

[Eghbal Hosseini @eghbal_hosseini]

Check out this fantastic review by @TalGolanNeuro, @SchuettHeiko, and @KriegeskorteLab! It is a tour de force on how to design optimal stimuli to study models of brain computation. highly recommend

[Ruxandra Teslo 🧬 @RuxandraTeslo]

Last week, when Anthropic announced the binder design news, I said the exciting part isn't binder design, but the orchestration of models to produce a research assistant - like entity. This is quite cool and it seems Anthropic is continuing down this road of scientific automation.

So what's happening?

They propose the Model Hardware Standard (MHS), which is a common standard that lets AI agents safely connect to and control physical scientific and industrial equipment. This would include things such as microscopes, liquid handlers, lasers and manufacturing machinery.

Today, integrating different pieces of hardware can take weeks or months because each device has its own interface and often requires custom software -- which is extremely annoying. MHS aims to reduce that setup to hours by giving devices a standardized software “driver” and a common way to communicate with AI agents.

The blog post gives one example, which involves Claude learning to align a laser. Claude adjusted the laser, then observed the result through a camera, then repeated the process to understand the system and finally, it wrote a script that could perform the alignment automatically.

I think this line of scientific automation
is actually much more exciting and higher yield potentially than all the stuff that aims more at making specific scientific discoveries (eg using AI to improve binder design). I think it's underrated how much boring friction slows down biological research, and it's exciting to potentially reduce that.

Of course, it also seems like being along the path to full scientific automation (including of human scientists) -- though still very early (and potentially it will never happen). I feel much more conflicted about that; I was just debating with some friends about this.

In my mind, scientific discovery is an expression of something very core to the human spirit -- a sort of fundamental curiosity. And I'd be sad if that ceased to be an activity humans were involved in.

[Lucas Beyer (bl16) @giffmana]

SWEs reviewing PRs:

[clem 🤗 @ClementDelangue]

BIG ANNOUNCEMENT FROM HUGGING FACE TODAY:

We're unveiling Microduck 🐥🤖

It's a tiny $399 open-source robot you can teach new tricks with reinforcement learning. It can walk, pick things up, get back up when it falls, and even roller-skate.

Welcome to the era of open-source affordable robots to democratize physical AI and world models!

🤗🤗🤗

[Avnish Kumar @avnish_ks]

🐐

[International Cyber Digest @IntCyberDigest]

Linus Torvalds says tracking down a bug in Linux's Intel GFX Driver was a "debug session from hell." The AI helping him repeatedly called it "impossible and unsolvable" and suggested they just write a report instead.

He kept pushing. 24 debug patches and 18 kernel boots later he found the bug: one round_up() that should have been round_down().

He then let the AI write the commit message.

https://t.co/9qcHPvThzP

[Jaana Dogan ヤナ ドガン @rakyll]

If you have any CPU nodes, hold on to them. I’m serious 😀

[Saagar Enjeti @esaagar]

Is there any good argument for why the Meta changes for teens shouldn't be universal for all adults?

[Jaana Dogan ヤナ ドガン @rakyll]

It's not unlikely that we will quickly get to a point where companies are hiring engineers solely to stop bad code from being merged.