iiinova — where your digital self keeps evolving

Stefan Gugler⚗️化学 @stevain.bsky.social · 28/06/2026

3/3 It was great fun to work with my collaborators at Xiamen Uni, @pavlodral.bsky.social and Yiyang Cui. rdcu.be/fqKLi

Architecture of Chinese chemical element names

Stefan Gugler⚗️化学 @stevain.bsky.social · 28/06/2026

2/3 About a year ago I asked myself how these aspects come together in Chinese characters and it turned out it's a convoluted history of nerds proposing and debating their ideas how a character for say Lithium should look and sound like.

rdcu.be

Architecture of Chinese chemical element names

121

Stefan Gugler⚗️化学 @stevain.bsky.social · 28/06/2026

A 'side project' of mine just got published in ChemTexts! (linked in last post) 1/3 I always really loved etymology and figuring out where certain words come from. The names of chemical elements are particularly interesting, as meaning, phonetics and societal stuff gets mashed together.

decomposition of the Chinese character for lithium into a semantic compound that means 'metal' and a phonetic compound that sound like 'li'

222

Stefan Gugler⚗️化学 @stevain.bsky.social · 30/03/2026

I was actually considering eos as well because of the recent manjaro drama ... what were your thoughts on why not plain arch instead of eos?

100

Reposted by Stefan Gugler⚗️化学

Felix Pultar @pultar.bsky.social · 12/01/2026

Ever wanted to run MD simulations of entire proteins in water with DFT accuracy? Meet AMPv3-BMS25, the latest iteration of our AMP multiscale neural network potential by @rinikerlab.bsky.social Read more in the preprints: doi.org/10.26434/che... doi.org/10.26434/che...

1145

Stefan Gugler⚗️化学 @stevain.bsky.social · 15/12/2025

I'll be at Pacifichem this week, hit me up if you wanna have a coffee! ☕

010

Reposted by Stefan Gugler⚗️化学

Adrian Hill @adrianhill.de · 03/12/2025

Join us today from 4:30 to 7:30 PM @neuripsconf.bsky.social Hall C,D,E #1006 for our poster on SmoothDiff, a novel XAI method leveraging automatic differentiation. 🧵1/6

1115

Stefan Gugler⚗️化学 @stevain.bsky.social · 02/12/2025

I'm in San Diego this week for Neurips, hit me up if you wanna have a coffee ☕

010

Stefan Gugler⚗️化学 @stevain.bsky.social · 22/06/2025

I'm in Oslo this week for WATOC, hit me up if you wanna have a coffee ☕

110

Stefan Gugler⚗️化学 @stevain.bsky.social · 12/03/2025

9/ Check it yourself: 🔗: github.com/khaledkah/tv... 📄: www.arxiv.org/abs/2502.08598 Thanks to Khaled, Winnie, Oliver, Klaus, and Shin for the cool collaboration as well as @bifold.berlin, TU Berlin, RIKEN, and DeepMind

github.com

GitHub - khaledkah/tv-snr-diffusion

Contribute to khaledkah/tv-snr-diffusion development by creating an account on GitHub.

021

Stefan Gugler⚗️化学 @stevain.bsky.social · 12/03/2025

8/ Takeaway: Exploding TV isn’t needed. Control TV + SNR separately for faster, better sampling. Method generalizes across domains (molecules, images).

110

Stefan Gugler⚗️化学 @stevain.bsky.social · 12/03/2025

7/ Why it works? Our empirical analysis shows: 1. Straight trajectories near data (t ≈ 0) are important (see in the inset plot) 2. Broad support of pₜ(𝐱) early on → robust to errors (note how SMLD goes from small to huge range instead of staying the same)

100

Stefan Gugler⚗️化学 @stevain.bsky.social · 12/03/2025

6/ Images: Matches EDM with uniform grid No fancy time grids like in EDM needed! VP-ISSNR on CIFAR-10/FFHQ ≈ EDM but with fewer hyperparameters!

100

Stefan Gugler⚗️化学 @stevain.bsky.social · 12/03/2025

5/ Molecules in 8 Steps: VP-ISSNR achieves 74% stability with 8 steps, 95% with 64 (SDE). Beats all baselines!

100

Stefan Gugler⚗️化学 @stevain.bsky.social · 12/03/2025

4/ We propose a new VP schedule 📈: Exponential inverse sigmoid SNR (ISSNR)→ rapid decay at start/end. Generalizes Optimal Transport Flow Matching.

110

Stefan Gugler⚗️化学 @stevain.bsky.social · 12/03/2025

3/ VP variants improve existing schedules: Take SMLD/EDM (exploding TV) → force TV=1. Result: +30% stability for molecules with 8 steps (x-axis is NFE=number of function evals).

100

Stefan Gugler⚗️化学 @stevain.bsky.social · 12/03/2025

2/ Most schedules (like EDM by Karras or SMLD (Song & Ermon) let TV explode (VE=variance exploding). We show constant TV (variance preserving, VP) + optimized SNR works better (ISSNR)! (it's a wild table, sorry, but notice our VP variants I circled)

100

Stefan Gugler⚗️化学 @stevain.bsky.social · 12/03/2025

1/ Problem: Diffusion models are slow due to repeated evals but reducing steps hurts quality if the noise schedule isn’t optimal. Other schedules passively adjust variance. Can we do better? 🔑Insight: control Total Variance (TV) and signal-to-noise-ratio (SNR) independently!

100

Stefan Gugler⚗️化学 @stevain.bsky.social · 12/03/2025

We have a new paper on diffusion!📄 Faster diffusion models with total variance/signal-to-noise ratio disentanglement! ⚡️ Our new work shows how to generate stable molecules in sometimes as little 8 steps and match EDM’s image quality with a uniform time grid. 🧵

151

Reposted by Stefan Gugler⚗️化学

CECAM @cecamevents.bsky.social · 05/03/2025

We are already on Day 3 of the workshop "Density Functional Theory and Artificial Intelligence learning from each other" in sunny CECAM-HQ. This afternoon, Fang Liu (Emory University) & Michael Herbst (EPFL) will present their talks.

061

Stefan Gugler⚗️化学 @stevain.bsky.social · 31/01/2025

our GPU cluster tonight after the ICML deadline

040

Reposted by Stefan Gugler⚗️化学

Adrian Hill @adrianhill.de · 30/01/2025

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

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.

314028

Stefan Gugler⚗️化学 @stevain.bsky.social · 31/01/2025

We submitted our stuff to the ICML today at lunch and I am so happy about it. It's juicy, it might get rejected, but it has the heart in the right place :]

130

Stefan Gugler⚗️化学 @stevain.bsky.social · 25/12/2024

acab includes the raclette police 💡

040

Stefan Gugler⚗️化学 @stevain.bsky.social · 17/12/2024

i guess their claim would be that it blows up for mysterious NN reasons rather than integrator or time step. 2 fs is a bit of chonky step, i agree, but if it explodes at say 0.1 fs i'd start wondering about the NN more than about the time step

100

Stefan Gugler⚗️化学 @stevain.bsky.social · 17/12/2024

stuff like this, (Ala)_2 at 2 fs or water at 1 fs? at <=0.5 fs they wouldn't explode for curl free forces, i assume?

100

Stefan Gugler⚗️化学 @stevain.bsky.social · 17/12/2024

why was that paper bad? i thought it was more of a benchmark than proposing their own thing anyways?

110

Stefan Gugler⚗️化学 @stevain.bsky.social · 20/11/2024

Seems like it refers to DFT. They actually give an intuition from metallurgy in the appendix about simulated annealing. I didn't know it's about removing impurities 🙉

000

Stefan Gugler⚗️化学 @stevain.bsky.social · 20/11/2024

Excellent question. Alas, I can't say i like mcmc (or tmcmc?) as a word either. A string of non-descript names is just ... Eugh

100

Stefan Gugler⚗️化学 @stevain.bsky.social · 20/11/2024

Ah yes, 'annealing', i do it every day and have a super intuitive understanding of what it is. In fact, im annealing right now.

100

Stefan Gugler⚗️化学 @stevain.bsky.social · 20/11/2024

Is "simulated annealing" really the best word for what it describes?

210

Stefan Gugler⚗️化学 @stevain.bsky.social · 20/11/2024

أيوةةة، حلو! أنا بتعلّم سوداني عربي.✨ ماشاءالله،يا يوليا!

110

Stefan Gugler⚗️化学 @stevain.bsky.social · 19/11/2024

The Pillar of Autumn (i know, i know)

020

Reposted by Stefan Gugler⚗️化学

Sophie Vershbow @svershbow.bsky.social · 19/11/2024

Week 2 actively using Bluesky

622815

Stefan Gugler⚗️化学 @stevain.bsky.social · 19/11/2024

in 10 years we will all be like 'remember when they made all these starter packs on bsky?'

020

Stefan Gugler⚗️化学 @stevain.bsky.social · 18/11/2024

One thing I can't square with the German culture of being on time is the always-late DB _sigh_

010

Stefan Gugler⚗️化学 @stevain.bsky.social · 18/11/2024

Train Berlin-Stuttgart was 6.5 h and juuuust bearable. Nice!

000

Stefan Gugler⚗️化学 @stevain.bsky.social · 17/11/2024

we wouldn't expect this to work since the train data is DFT which also doesn't get Cr2 dissociation correct, right?

000

Stefan Gugler⚗️化学 @stevain.bsky.social · 17/11/2024

a lot of new connections due to the chemistry lists going around ... it's cool but also let's not make it too linkedin-y with the professionalism

020

Stefan Gugler⚗️化学 @stevain.bsky.social · 16/11/2024

while gpt helps a lot with very technical writing, if the text should have any wit, any juice, anything reminiscent of prose, it's even more useless than i thought.

000

Stefan Gugler⚗️化学 @stevain.bsky.social · 15/11/2024

im writing a pop sci article and forgot how difficult it is, uff

000

Stefan Gugler⚗️化学 @stevain.bsky.social · 09/02/2024

and they're all a threatening 6 foot 6 T_T

000

Reposted by Stefan Gugler⚗️化学

👁️👅👁️ @alisonbuki.bsky.social · 05/02/2024

just one more simulacrum bro. i promise bro just one more simulacrum and it'll fix everything bro. bro... just one more simulacrum. please just one more. one more simulacrum and we can fix this whole problem bro. bro c mon just give me one more simulacrum i promise bro. just one more and hypereality will be perfect bro. bro bro please i just need one more simulacrum

072