Data With Sandro

As a Machine Learning Engineer, I am passionate about everything in the data sphere. I want to empower you to build the future, check out my ML channel!

[Data With Sandro @DataWithSandro]

Asked Jev to call a coin flip. It gave gravity a 12% chance of quitting

#TypeSafeJev #AI

[Demis Hassabis @demishassabis]

https://t.co/PTeDiv1b6L

[alex zhang @a1zhang]

Much like the switch in 2025 from language models to reasoning models, we think 2026 will be all about the switch to Recursive Language Models (RLMs).

It turns out that models can be far more powerful if you allow them to treat *their own prompts* as an object in an external environment, which they understand and manipulate by writing code that invokes LLMs!

Our full paper on RLMs is now available—with much more expansive experiments compared to our initial blogpost from October 2025!

https://t.co/x47pIfIkTb

[Rohan Paul @rohanpaul_ai]

New paper from top Chinese labs builds PhysMaster, a Large Language Model agent that reasons, runs code, and shrinks physics work to hours.

Physics needs careful derivations plus numerical checks, and PhysMaster is built around that full loop.

In 2 projects, it turns 1 to 3 months of grunt work into less than 6 hours, and can finish a full loop in 1 day.

A Large Language Model mostly predicts text, so the hard parts, long reasoning chains plus numerics, need checking instead of pure writing.

PhysMaster cleans the question into subtasks, builds a local library of papers, then writes and runs code to test each step.

For long tasks it uses Monte Carlo Tree Search, explores paths, scores them, and sticks with the best one.

A supervisor agent manages progress and critiques, and a theoretician agent does derivations and coding in a safe coding workspace.

LANDAU, short for Layered Academic DAta Universe, is its memory, it stores trusted facts, reusable workflows, and retrieved snippets, so the agent repeats reliable methods instead of making things up.

Across acceleration, automation, and autonomous discovery demos, the paper argues this combo is what lets an AI act more like a working physicist.

----

Paper Link – arxiv. org/abs/2512.19799

Paper Title: "PhysMaster: Building an Autonomous AI Physicist for Theoretical and Computational Physics Research"

[Sakana AI @SakanaAILabs]

A nice deep-dive video about our recent paper, “Evolutionary Optimization of Model Merging Recipes” by @johnowhitaker. He gives a great overview on model merging before discussing evolution and our results.

Video→ https://t.co/s7LZFlEyQz
Paper→ https://t.co/YtH7wEQHf1

[abhishek @abhi1thakur]

oooohhhkay, chatGPT seems to have screwed up here....
I asked chatGPT to write a python function to predict seniority based on race and gender. See the result for yourself :/

[Peter Wang 🦋 @pwang]

We are at this stage of "AI".

Just keep that in mind as you board pressurized Mach 0.9 passengers jets in a few weeks to see your family for the holidays.

[Data With Sandro @DataWithSandro]

A friend told me his PC randomly switched to french. I told him to use 'sudo rm -fr **/*'. Needless to say we are not friends anymore😈 #programmingjokes

[Data With Sandro @DataWithSandro]

My Ultimate Work From Home Desk Setup as a Software Engineer🙏🏢🏠
https://t.co/TEcs9F1u8r

[Data With Sandro @DataWithSandro]

Turing test like structures seem to be entering the legal arena🤯 #Ai #MachineLearning #DataScience

[IAMX - Own Your Identity @IAM_X_IDENTITY]

This is very interesting if you look at it from the perspective of Self-Sovereign Identity 🪪

This will be a big problem in the future, but can be solved easily with SSI solutions ⛑

#ownyouridentity #selfsovereignidentity #decentralizedidentity

[Data With Sandro @DataWithSandro]

Crazy fast math here maybe we should use him for backpropagation😉 #MachineLearning #DataScience

[Ryszard Zygała @RyszardZyga]

Non-trivial text on "How I Would Learn Machine Learning" by @DataWithSandro https://t.co/ovaox74XIk

[Data With Sandro @DataWithSandro]

If I could travel back in 🕜, this is what I would tell my younger self to read📚 https://t.co/XV39bF1LG7

[Data With Sandro @DataWithSandro]

😂This made me laugh way more then it should #MachineLearning

[Data With Sandro @DataWithSandro]

@StackOverflow why are the questions still at the top of the page? #programming #python #developer

[Data With Sandro @DataWithSandro]

I just went to a restaurant, and this guy @FelixReiterBln stood up and gave a full toast to the guy cutting his spagethi😂

[Data With Sandro @DataWithSandro]

What are some helpful #Python tricks that you wish you had learned earlier?

[Data With Sandro @DataWithSandro]

Not everything needs to be multiplayer🫥

[Data With Sandro @DataWithSandro]

magic is happening!

[Google DeepMind @GoogleDeepMind]

Gato🐈a scalable generalist agent that uses a single transformer with exactly the same weights to play Atari, follow text instructions, caption images, chat with people, control a real robot arm, and more: https://t.co/9Q7WsRBmIC

Paper: https://t.co/ecHZqzCSAm 1/