vishal

Data Science/ML.
Maintainer.
Go Birds. #FlyEaglesFly

[vishal @vishal_learner]

in a previous blog post, I used the Anthropic Python SDK with Sonnet 4.6 to create a trivial proof-of-concept data analysis agent with custom tools. In this blog post, I use Qwen3-1.7B to achieve the same result!!

https://t.co/5kQe0VP7BC

[vishal @vishal_learner]

i've been thinking about the concept of "LLM as an interface" for awhile and after getting some experience building some AI workflows for projects in prod, i have put together my thoughts in this blog post:

https://t.co/6j6ZxSv4fS

[Neek @babyltaly]

i fear the Eagles winning another SuperBowl

[vishal @vishal_learner]

@howmanysmaII @TragicWarrior Exactly but my concern was if this is a killed-by-Google trajectory, Bryan's logic makes more sense. Especially if there's a shift towards more open source usage.

[vishal @vishal_learner]

im open to being convinced that this is a good thing

[Andrew Curran @AndrewCurran_]

Well that escalated quickly. The Information is reporting that the deal is done.

[Honest NFL @TheHonestNFL]

Mourning and honoring Dolly Parton may be the only thing the entire galaxy can agree on right now

[vishal @vishal_learner]

@ShaneHaffNFL Is it just me or do those unis make everyone in that pic look shorter? Is it the camera?

[vishal @vishal_learner]

@iamtrask Can you elaborate on what the major market correction is in your estimation?

[vishal @vishal_learner]

So I know I'm out of my depth here with this topic and I understand maybe 60% of these pages but isn't macro behavior emergent and not just an aggregate of sub-atomic behavior? u.e. what's true for photons isn't necessarily true at human scale?

[vishal @vishal_learner]

one topic I've been fascinated about is the relationship b/w causality and time. I've had a many hours of conversations with Claude over the past year on this topic, helping me find existing literature that either supports or refutes my thinking...

[Chris Albon @chrisalbon]

Nvidiaface

[vishal @vishal_learner]

@TragicWarrior This makes sense thank you. I like that.

[tomaarsen @tomaarsen]

Which checkpoint you start from matters more than I expected, and the intuitive choice is wrong.

The -unsupervised checkpoints adapt far better than their finished siblings:

mLateOn-unsupervised: +3.1 points
mLateOn: +0.4 points

[Andrew Curran @AndrewCurran_]

Well that escalated quickly. The Information is reporting that the deal is done.

[Eagles Nation @PHLEaglesNation]

Jalen Hurts: “Experiences don’t mean anything until you reflect on them, grow from them, and start to apply them.”

#Eagles

[vishal @vishal_learner]

https://t.co/PdaL71ajoV

[vishal @vishal_learner]

one topic I've been fascinated about is the relationship b/w causality and time. I've had a many hours of conversations with Claude over the past year on this topic, helping me find existing literature that either supports or refutes my thinking...

[vishal @vishal_learner]

one topic I've been fascinated about is the relationship b/w causality and time. I've had a many hours of conversations with Claude over the past year on this topic, helping me find existing literature that either supports or refutes my thinking...

https://t.co/PdaL71ajoV

[vishal @vishal_learner]

https://t.co/PdaL71ajoV

[vishal @vishal_learner]

been calling "saving intermediate artifacts before/after transforms in a pipeline so you can see if it's behaving the way it should" as "observability" but implementing the code is technically "instrumentation" + "observability" seems to be about inferring not direct observation

[Fav @Favwontmiss]

My daughter asked me what games I played on the tablet when I was a kid, and I told her I used to yell into a fan to sound like a robot.

[Kareem Carr, Ph.D. @kareem_carr]

This is an incredible time. After 1000+ years, math is becoming a semi-empirical science where you understand things to be correct because they have been computer verified, and not because you personally understand them from first principles.

[Ash Jogalekar @curiouswavefn]

I am guilty of this myself. I am about to submit an AI-derived mathematical proof to arXiv that I don't completely understand but which addresses a problem in theoretical biology with a neat solution. I understand the problem, the approach and the result, but I don't understand all the details of the proof. It's been an exciting experience that has occupied the better part of the last month for me. I am transparent about my limitations in understanding the proof and welcome human review that proves it right or wrong (you learn something important about AI in either case).

In general, I consider it an entirely good thing for people who are not specialized to use AI for solving problems in their field using tools from other fields that they don't fully understand. But I agree that it is placing a serious burden on human verifiers. Unless, of course, the AIs become so good at verifying that we don't need humans.