Scholar
LightNews

John Gardner

Videos
H-index: 31
Political science 23%
Engineering 21%
more

by John Gardner

jla-gardner.bsky.social
Thanks! 😊 In principle yes - our data generation protocol requires ~5 model calls to generate a new, chemically reasonable structure + is easily parallelised across processes. If you were willing to burn $$$ one could generate a new dataset very quickly (as is often the case with e.g. RSS set ups)
3

by John Gardner — Reposted by John Gardner

jla-gardner.bsky.social
Excited to share the pre-print we’ve been working on for the last ~4 months:

“Distillation of atomistic foundation models across architectures and chemical domains”

Deep dive thread below! 🤿🧵
3 2 11

by John Gardner

jla-gardner.bsky.social
... @juraskova-ver.bsky.social, Louise Rosset, @fjduarte.bsky.social, Fausto Martelli, Chris Pickard and @vlderinger.bsky.social 🤩
1

by John Gardner

jla-gardner.bsky.social
It was super fun collaborating with my co-first-author @dft-dutoit.bsky.social, together with the rest of the team across various research groups: @bm-chiheb.bsky.social, Zoé Faure Beaulieu, Bianca Pasça...
1

by John Gardner

jla-gardner.bsky.social
We hope that you can start using this method to do cool new science!
1

by John Gardner

jla-gardner.bsky.social
Code for our synthetic generation pipeline (compatible with any ase Calculator object) can be found here:
github.com/jla-gardner/...
GitHub - jla-gardner/augment-atoms: dataset augmentation for atomistic machine learning
dataset augmentation for atomistic machine learning - jla-gardner/augment-atoms
github.com
1

by John Gardner

jla-gardner.bsky.social
For (many) more results, please see the pre-print! arxiv.org/abs/2506.10956
Distillation of atomistic foundation models across architectures and chemical domains
Machine-learned interatomic potentials have transformed computational research in the physical sciences. Recent atomistic `foundation' models have changed the field yet again: trained on many differen...
arxiv.org
1

by John Gardner

jla-gardner.bsky.social
I find our results for the modelling of MAPI (a hybrid perovskite) particularly pleasing: the distribution of cation orientations generated by the teacher and student models during NVT MD are ~identical!
1

by John Gardner

jla-gardner.bsky.social
We go on to apply this distillation approach to target other chemical domains by distilling different foundation models (Orb, MatterSim @msftresearch.bsky.social) and MACE-OFF), and find that it works well across the board!
1

by John Gardner

jla-gardner.bsky.social
Beyond error metrics, we extensively validate these models to show they model liquid water well.
1

by John Gardner

jla-gardner.bsky.social
These student models have relatively few parameters (c. 40k for PaiNN and TensorNet), and so have much lower memory footprint. This lets you scale single GPU experiments very easily!
1

by John Gardner

jla-gardner.bsky.social
The resulting student models reach impressive accuracy vs DFT while being orders of magnitude faster than the teacher!

Note that these student models are of a different architecture to MACE, and in fact ACE is not even NN-based.
1

by John Gardner

jla-gardner.bsky.social
We start by (i) fine-tuning MACE-MP-0 (@ilyesbatatia.bsky.social) on 25 water structures labelled with an accurate functional, (ii) using this fine-tuned model and structures to generate a large number (10k) new “synthetic” structures, and (iii) training student models on this dataset.
1

by John Gardner

jla-gardner.bsky.social
Does this distillation approach work? In short, yes! 🤩
1

by John Gardner

jla-gardner.bsky.social
This approach is very cheap, taking c. 5 calls to the teacher model to generate a new, chemically relevant and uncorrelated structure! We can build large datasets within one hour using this protocol.
1

by John Gardner

jla-gardner.bsky.social
In this pre-print, we propose a different solution: starting from a (very) small pool of structures, and repeatedly (i) rattling and (ii) crudely relaxing them using the teacher model and a Robbins-Monro procedure.
1

by John Gardner

jla-gardner.bsky.social
This works well, but has two drawbacks: (1) MD is still quite expensive, and requires many steps to generate uncorrelated structures, and (2) expert knowledge and lots of fiddling is required to get the MD settings right.
1

by John Gardner

jla-gardner.bsky.social
In previous work, we and others (PFD-kit) have proposed using teacher models to generate "synthetic data" by using them to drive MD, and to sample snapshots along these trajectories as training points.
1

by John Gardner

jla-gardner.bsky.social
The devil is always in the details however 😈 The main problem we need to solve is how to generate many relevant structures that densely sample the chemical domain we are interested in targeting.
1

by John Gardner

jla-gardner.bsky.social
At a high level, this builds upon the approach pioneered by Joe Morrow, now extended to the distillation of impressively capable foundation models, and to a range of downstream architectures and chemical domains,
1

by John Gardner

jla-gardner.bsky.social
Concretely, we train a student to predict the energy and force labels generated by the teacher on a large dataset of structures: this requires no alterations to existing training pipelines, and so is completely agnostic to the architecture of both the teacher and student 😎
1

by John Gardner

jla-gardner.bsky.social
Both of the above methods try to maximise the amount of information extracted per training structure from the teacher. Our approach is orthogonal to this: we try to maximise the number of structures (that are both sensible and useful) we use to transfer knowledge.
1

by John Gardner

jla-gardner.bsky.social
Somewhat similarly,
@ask1729.bsky.social
and others extract additional Hessian information from the teacher. Again, this works well providing you have a training framework that lets you train student models on this data.
1

by John Gardner

jla-gardner.bsky.social
@gasteigerjo.bsky.social
and others attempt to align not only the predictions, but also the internal representations of the teacher and the student. This approach works well for models with similar architectures, but is incompatible with e.g. fast linear models like ACE.
1

by John Gardner

jla-gardner.bsky.social
At their heart, all model distillation strategies attempt to extract as much information as possible from the teacher model, in a format that is useful for the student.

Various existing methods in the literature do this in different ways.
1

by John Gardner

jla-gardner.bsky.social
In the context of machine learned interatomic potentials, distillation lets you simulate larger systems, for longer times, and with less compute.
This lets you explore new science, and democratises access to otherwise expensive simulations/methods and foundation models. 💪
1

by John Gardner

jla-gardner.bsky.social
Model distillation methods take a large, slow “teacher model”, and try to condense, or “distill”, its knowledge into a smaller, faster “student model”

If this can be done well, it is an extremely useful thing!
1

by John Gardner

profjohngardner.bsky.social
'da, in my part of the world, means father'--Samuel Beckett
4

by John Gardner

jla-gardner.bsky.social
If you want to dive straight in with some hands-on examples, you can find several links to Colab notebooks in the graph-pes docs:
jla-gardner.github.io/graph-pes/
graph-pesContentsMenuExpandLight modeDark modeAuto light/dark, in light modeAuto light/dark, in dark mode
jla-gardner.github.io

by John Gardner

jla-gardner.bsky.social
If graph-pes sounds interesting to you, please do check out the repo and give it a star!
Please also reach out via GitHub issues or DM on here if you have any questions or feedback.
github.com/jla-gardner/...
GitHub - jla-gardner/graph-pes: train and use graph-based ML models of potential energy surfaces
train and use graph-based ML models of potential energy surfaces - jla-gardner/graph-pes
github.com
1 1

References

Fields & subjects

Updated 1m