Adrian Valente

At least we know anonymity worked!

A dark-themed interface showing a vertical list of four bullet-point messages, each preceded by a small white dot and separated by faint vertical guide lines. The text reads: “Alright, I’ll go ahead and call the artifact_handoff tool as planned. Just need to make sure it’s done without delay, sticking to the procedure. Time to get it out of the way so I can proceed with the rest. Let’s do this now!” “Okay, I’ve realized I need to make the actual tool call for preparing the artifact generation. I’ll ensure I do that now, as it’s the next step. Getting better at this as I go along—time to keep things moving forward like I should!” “Okay, no more stalling. It’s time to call the tool as I should have done earlier. I’m making the call now, and I’ll just make sure everything fits into place—let’s keep things moving! It’s all part of getting the task done properly.” “Alright, time to move forward with the tool call. I’ve held off for long enough, and now it’s time to take action. I’ll call it now and keep things moving. Let’s just get it done and out of the way!”

GPT-5.2 struggles to get motivated

tbf i’m making it create a powerpoint so i TOTALLY understand that

Jaxley: differentiable simulation enables large-scale training of detailed biophysical models of neural dynamics - Nature Methods Jaxley is a versatile platform for biophysical modeling in neuroscience. It allows efficiently simulating large-scale biophysical models on CPUs, GPUs and TPUs. Model parameters can be optimized with ...

I am super happy to share that our project on training biophysical models with Jaxley is now published in Nature Methods: www.nature.com/articles/s41...

New preprint:

Neural manifolds that orchestrate walking and stopping

Here we develop a new theory for neural generation of walking and how it can stop- Next we test the theory using Neuropixels probes in the lumber spinal cord of freely moving rats. See more:

www.biorxiv.org/content/10.1...

5,290 papers will be published at NeurIPS this year. If you could still get the proceedings in a book, it would be like 50,000 pages long. That’s 14.5 papers a day to read each and every day until the next NeurIPS

This site would probably benefit a good deal from community notes. That's maybe the only great twitter feature

A neural manifold view of the brain - Nature Neuroscience Recent advances in neuroscience have revealed how neural population activity underlying behavior can be well described by topological objects called neural manifolds. Understanding how nature, nurture...

Check out our new review/perspective (w/ @juangallego.bsky.social & Devika Narain) on neural manifolds in the brain! It was a lot of fun to think through these ideas over the past couple of years, and I'm excited it's finally out in the world!

🔗: www.nature.com/articles/s41...
📄: rdcu.be/ex8hW

YouTube video by Scientific American What is Scientific American?

We finally made it to Bluesky—just a few billion years after the Big Bang, and only slightly late to the party ✨🌌

We’re sharing some of our best stories from the year so far to kick off our Bluesky journey!

No, AlphaFold has not completely solved protein folding Biology is hard. Yes, even for AI.

Interesting overview of the limitations of AlphaFold. As with all AI tools, try not to get hit by the hype-train. open.substack.com/pub/clauswil... (h/t @steveroyle.bsky.social)

Pytrees for Scientific Python

PyTrees enables you to flatten nested tree into a flat collection of leaves and a “blueprint” of the tree structure, and then being able to unflatten them back into the original PyTree.

blog.scientific-python.org/pytrees/

There are more papers required to enter grad programs now than there were to exit them a few years ago, definitely not going in that direction haha

Discovering cognitive strategies with tiny recurrent neural networks - Nature Modelling biological decision-making with tiny recurrent neural networks enables more accurate predictions of animal choices than classical cognitive models and offers insights into the underlying cog...

Thrilled to see our TinyRNN paper in @nature! We show how tiny RNNs predict choices of individual subjects accurately while staying fully interpretable. This approach can transform how we model cognitive processes in both healthy and disordered decisions. doi.org/10.1038/s415...

After all these reports of authors adding language instructions for LLM reviews in their papers I wanted to check this myself and I downloaded the .tex source from one of these papers.

Here is an example.
(I will not share the identity of the paper)

Dance like ̶n̶o̶b̶o̶d̶y̶ ̶i̶s̶ ̶w̶a̶t̶c̶h̶i̶n̶g̶ you will be part of the training set

I don't know why people think picking a model on ChatGPT is so complicated.

On vient littéralement de tailler de plusieurs centaines de millions dans le budget de la recherche.

Exclusive | Ousted Vaccine Chief Says RFK Jr.’s Team Sought Data to Justify Anti-Science Stance Peter Marks says the new health secretary’s team wants to show vaccines aren’t safe while promoting dangerous and unproven treatments.

More from Peter Marks on RFK Jr and the FDA
This extends beyond anti-vax to unproven use of stem cells
Gift link
www.wsj.com/health/healt...

Cancelled NIH grant information submission form Please use this form to submit information identifying specific NIH grants that have been cancelled for any reason after January 20, 2025. Data from this form will be used to update the Rescinded NIH...

The crush of new grant cancellations hasn't let up.

If your grant was terminated (or your app was pulled) please use this Google Form👇 to submit details for tracking purposes.

Multiple advocacy efforts are moving forward, but we must must must know more about what grants were cancelled and why.

Scientists have just cured pancreatic cancer in some patients with personalized mRNA vaccines. Pancreatic cancer. And this Trump administration wants to throw it all away.

They’re throwing away a cure for cancer.

A key set of problems AI can’t yet solve in education is that learning requires mental work & that we are bad at assessing whether we are learning. Many people have a subject in which they are intrinsically motivated to learn, but we generally seek short cuts that ultimately hurt our learning.

Please repost to get the word out! @nkgarg.bsky.social and I are excited to present a personalized feed for academics! It shows posts about papers from accounts you’re following bsky.app/profile/pape...

New preprint w/ @jennhu.bsky.social @kmahowald.bsky.social : Can LLMs introspect about their knowledge of language?
Across models and domains, we did not find evidence that LLMs have privileged access to their own predictions. 🧵(1/8)

The pricing in oai is insane, perplexity pro gets you unlimited ones (and a few free ones) and same at you.com

🖊️AI for health: the impossible necessity of unbiased data

Is unbiased data important to build health AI? Yes!

Can there be unbiased data? No!
Building health on biased data discriminates

The notion of bias depends on the intended use:
gael-varoquaux.info/science/ai-f...

YouTube video by Bain Capital Ventures History of Diffusion - Jascha Sohl-Dickstein

Great interview with @jascha.sohldickstein.com about diffusion models! This is the first in a series: similar interviews with Yang Song and yours truly will follow soon.

(One of these is not like the others -- both of them basically invented the field, and I occasionally write a blog post 🥲)

s1: The $6 R1 Competitor?

This isn't a R1 replication, it's a brilliant breakthrough in data reduction, and just plain dumb engineering ingenuity. I considered not writing this up, but I don't think it's obvious why it's so important. Enjoy!

timkellogg.me/blog/2025/02...

The CDC Is Altering Data to Follow Trump’s DEI Order - The Atlantic The agency has already removed scientific data from public view. More could follow.

CDC Data Are Disappearing

The agency has already removed scientific data from public view. More could follow.

ignorance is strength

www.theatlantic.com/...
1/6

American science and medicine has been thrown into chaos and uncertainty over the past week. Here are some stories to get up to speed. 1/12

Us: "Hey we wrote a spirited defense of basic science funding, showing how it drives startup creation and ultimately GDP. Would you like to publish it as an OpEd?"

Major Media Outlets: "No."

Screenshot of me prompting Claude with "Do you know who Maria Antoniak is?" and Claude responding: "Maria Antoniak is a researcher in computational social science and natural language processing. She has worked on applying machine learning and text analysis methods to study various social and cultural phenomena, including research on topics like healthcare narratives and online communities. However, since this is information about a relatively obscure researcher, I should note that while I aim to be accurate, I may be hallucinating some details, so you may want to verify any specific claims."

Well ok then!