Ben Miller

GitHub - facebookresearch/adjoint_sampling: code for "Adjoint Sampling: Highly Scalable Diffusion Samplers via Adjoint Matching" code for "Adjoint Sampling: Highly Scalable Diffusion Samplers via Adjoint Matching" - facebookresearch/adjoint_sampling

We've open sourced Adjoint Sampling!

It's part of a bundled release showcasing FAIR's research and open source commitment to AI for science.

github.com/facebookrese...

There's also a blog about the release
ai.meta.com/blog/meta-fa...

This was a significant community effort, so thank you to LBNL, Princeton, Genentech, Stanford, U. of Cambridge, CMU, NYU, LANL, the Roche Group and Berkeley for helping make this happen!

UMA, without fine-tuning to specific tasks, performs similarly or better in both accuracy and inference-speed/memory-efficiency than specialized models on a wide-range of material, molecular and catalysis benchmarks.

With all these large diverse datasets, we ask whether it’s possible to train a single model that performs well across several domains without the need for additional finetuning?

Introducing UMA - our latest model trained on over 30 billion atoms across our open-science datasets.

Baseline models are released and evaluated on OOD evaluation tasks. Metrics that go beyond just a structure’s energy and forces are important for understanding how well models perform on downstream applications.

We hope to release a public leaderboard in the near future!

With over 100M structures calculated at ωB97M-V/def2-TZVPD, with variable charge+spin, and spanning biomolecules, electrolytes, metal complexes, and community datasets, OMol25 represents the largest most diverse dataset of its kind.

While our earlier datasets (OC20, OC22, OMat24) focused on catalysts and materials, it was evident the community lacked a large, high quality, diverse molecular DFT dataset for training ML models. The Open Molecules 2025 (OMol25) Dataset hopes to accomplish just that.

Excited to share our latest releases to the FAIR Chemistry’s family of open datasets and models: OMol25 and UMA!
@AIatMeta @OpenCatalyst

OMol25: huggingface.co/facebook/OMo...
UMA: huggingface.co/facebook/UMA
Blog: ai.meta.com/blog/meta-fa...
Demo: huggingface.co/spaces/faceb...

Any take aways?

Really happy with this work led by Aaron! Take a look people!

@franknoe.bsky.social @wellingmax.bsky.social

What did they do?

Excited to be a keynote speaker at the Netherlands Society on Biomolecule Modelling annual meeting tomorrow!

I'll be speaking about All Atom Diffusion Transformers for the first time publicly!

Thank you @erikjbekkers.bsky.social for the invitation 🙏

www.nsbm.nl

Meet @ivonajdenkoska.bsky.social, Postdoc Researcher at the Uni Amsterdam 🇳🇱. She researches multimodal #FoundationModels, developing efficient methods for vision-language understanding and generative tasks. Learn a career highlight of hers and her advice for young scientists. #WomenInELLIS #AI #ML

God, that was hard to bear. Students told a physics town hall that grad schools have rescinded admission. NSF proposals are censored for words like “bias,” even when the meaning is technical. Physics curricula are being banned by school districts for encouraging girls. #APSGlobalSummit2025 #APSMarch

Chaitanya led a very impressive paper here! Congrats and yay for unified representations!

10 maart: Universiteit Leiden start estafettestaking tegen bezuinigingen Medewerkers van de Universiteit Leiden trappen maandag 10 maart de estafettestaking af van de Nederlandse universiteiten tegen de kabinetsbezuinigingen.

Here is tomorrow’s program. We’re meeting at 11 a.m. at the Garenmarktplein in Leiden. The program starts at 11.30. The teach outs are throughout the day in The Hague and Leiden: www.universiteitleiden.nl/en/news/2025...

Looks cool!

We are experiencing an assault on science unparalleled by anything I’ve seen in my life. It’s not one issue or another anymore, the entire institution is under attack by the most powerful individuals in the country.

This Friday, where will you be?

standupforscience2025.org

Looks very cool!

ELLIS Welcomes InvestAI Initiative as a Strong Endorsement of Its Mission The ELLIS mission is to create a diverse European network that promotes research excellence and advances breakthroughs in AI, as well as a pan-European PhD program to educate the next generation of AI...

ELLIS welcomes the InvestAI initiative as a strong step toward strengthening AI research, computing infrastructure, and talent development in Europe.

Read our full statement: ellis.eu/news/ellis-w...

#AIActionSummit

Get in dorks, we're going protesting!

STAND UP FOR SCIENCE WITH US ON MARCH 7TH, 2025
WASHINGTON DC AND EVERY STATE CAPITOL

Because science is for everyone!

Find us at standupforscience2025.org

#standupforscience2025 #scienceforall #sciencenotsilence

Doesn't the timing of this not make sense?

Figure comparing automatic differentiation (AD) and automatic sparse differentiation (ASD). (a) Given a function f, AD backends return a function computing vector-Jacobian products (VJPs). (b) Standard AD computes Jacobians row-by-row by evaluating VJPs with all standard basis vectors. (c) ASD reduces the number of VJP evaluations by first detecting a sparsity pattern of non-zero values, coloring orthogonal rows in the pattern and simultaneously evaluating VJPs of orthogonal rows. The concepts shown in this figure directly translate to forward-mode, which computes Jacobians column-by-column instead of row-by-row.

You think Jacobian and Hessian matrices are prohibitively expensive to compute on your problem? Our latest preprint with @gdalle.bsky.social might change your mind!
arxiv.org/abs/2501.17737
🧵1/8

If you’re in the US and you’d like to know what projects and vital services federal grants currently fund in your state, you can search here: www.usaspending.gov

And you can find the contact information for your elected representatives here: www.usa.gov/elected-offi...

They need to hear from you.

🙏 Please help us improve the SBI toolbox! 🙏

In preparation for the upcoming SBI Hackathon, we’re running a user study to learn what you like, what we can improve, and how we can grow.

👉 Please share your thoughts here: forms.gle/foHK7myV2oaK...

Your input will make a big difference—thank you! 🙌

USDOT has now deleted this webpage

Looks beautiful!

I have heard it has some weird sharp edges still though... what do you think? Have you replaced latex entirely?

The template for this poster is now available on GitHub!

github.com/francois-roz...

Trump hits NIH with ‘devastating’ freezes on meetings, travel, communications, and hiring Researchers facing

The NIH actions go far beyond the DEI crackdown. All external advisory meetings are canceled, in ones that were in progress today. That means that NIH cannot issue new grants, as external review is required by law. It's a big deal. See, e.g. www.science.org/content/arti...

I sadly won't attend (at least not in person) this year, but it was great fun with great people!