---
title: 'Universal Atomistic Models: UMA Overview'
url: https://www.emergentmind.com/topics/universal-models-for-atoms-uma
type: topic
---

# Universal Atomistic Models: UMA Overview

Searching arXiv for relevant UMA and precursor papers to ground the article in current literature.
Universal Models for Atoms (UMA) denotes a family of universal, pretrained graph-based machine-learned interatomic potentials (MLIPs) developed as general-purpose atomistic foundation models for molecules, materials, catalysts, molecular crystals, and metal–organic frameworks. In the modern formulation, UMA is designed to test whether a single universal model can achieve strong zero-shot performance across these domains while remaining fast enough for practical simulation workloads such as molecular dynamics, relaxations, and structure search; the central claim is that a single model without any fine-tuning can perform similarly or better than specialized models [2506.23971]. In a broader historical sense, the UMA agenda also includes earlier atom-centered and local-environment-centered approaches that treated transferable atomic environments, rather than whole molecules or crystals, as the basic learning unit, most notably the AML/amons framework [1707.04146].

## 1. Conceptual basis of universal atomistic modeling

A recurring idea across UMA-style work is that atomistic prediction can be organized around transferable local environments. In the AML/amons framework, the total energy is written as a sum of atomic contributions,
$$
E=\sum_I E^I,
$$
and the machine-learning model is posed at the atomic level,
$$
E^I-\varepsilon_0^I=\phi(\mathbf{M}^I)^\top \mathbf{w}+\epsilon.
$$
In the Gaussian-process/KRR formulation, molecular covariance is decomposed into atomic covariances,
$$
K_{ij}=\sum_I\sum_J K_{ij}^{IJ}, \qquad K_{ij}^{IJ}=k(\mathbf{M}_i^I,\mathbf{M}_j^J),
$$
with a Gaussian kernel and atom-type matching. This yields a transferable mapping from local atomic environments to property contributions rather than a monolithic molecular map [1707.04146].

The same paper defines amons as atoms-in-molecules-based fragments selected on-the-fly from a query system. The fragment-selection rule is graph-theoretic and chemically constrained: connected subgraphs are enumerated, the original hybridization and valence state must be preserved, and exact subgraph isomorphism to the query graph is required. For noncovalent systems the procedure is adapted with vdW bonds and special selection rules, while for very large biomolecules static amons are obtained as exact cutouts around a target atom within a cutoff radius. This suggests that one early route to universality was not a single end-to-end universal potential, but an atomwise reconstruction strategy based on a finite dictionary of recurring local environments [1707.04146].

A complementary precursor is the Gaussian multipole (GMP) featurization, which was introduced to avoid element-specific feature growth. GMP encodes chemical identity through an approximate electron density built from atom-centered Gaussians and probes that density with radial Gaussians and Maxwell–Cartesian spherical harmonics. The basic feature is
$$
\mu_{i,abc}=\langle S_{abc}\times G_i,\hat{\rho}\rangle,
$$
and rotational invariance is imposed through grouped norms over permutation-related harmonics. The resulting feature vector has fixed dimension regardless of the number of elements present, enabling a single neural network rather than separate subnetworks per element type. The authors explicitly position this as a route toward a more universal atomistic model whose input representation can be reused across many element types, system classes, and datasets [2102.02390].

## 2. UMA architecture, model family, and scaling laws

In its current form, UMA is built on eSEN, an equivariant graph neural network, with the principal architectural novelty given by the Mixture of Linear Experts (MoLE) mechanism. The standard expert combination is
$$
y=\sum_k \alpha_k (W_k x),
$$
which UMA rewrites as
$$
y=W^*x,\qquad W^*=\sum_k \alpha_k W_k.
$$
Because the expert mixture is computed once from global, time-invariant information and merged before the forward pass, the model gains large total parameter count without a proportional inference penalty. The inputs therefore include not only atom positions and atomic numbers, but also global system descriptors such as charge, spin multiplicity, and the DFT task label [2506.23971].

The published UMA family contains three principal variants. UMA-S has **150M total parameters**, **6M active parameters**, about **16 inferences/sec for 1k atoms**, and fits **100k+ atoms** on an 80GB GPU. UMA-M has **1.4B total parameters**, **50M active parameters**, about **3 inferences/sec for 1k atoms**, and fits **10k+ atoms** on an 80GB GPU. UMA-L has **700M total parameters**, **700M active parameters**, about **1.6 inferences/sec for 1k atoms**, and fits **1k+ atoms** on an 80GB GPU. On a typical H100 benchmark with approximately 50 neighbors per atom, UMA-S can simulate **1000 atoms at 16 steps/sec**, corresponding to about **1.4 ns/day** [2506.23971].

A central technical contribution is the empirical study of scaling laws at atomistic-model scale. Training compute is modeled as
$$
C(N,D)\approx \kappa N D,
$$
with $\kappa\approx 270$ FLOPs/parameter/atom for the UMA-M-like setting. Compute-optimal size laws are then fit as
$$
\log(N^*(C))=\alpha\log(C)+A,\qquad \log(D^*(C))=\beta\log(C)+B,
$$
and the usual two-variable loss ansatz is written as
$$
\tilde{L}(N,D)=\hat{E}+\frac{\hat{A}}{N^{\hat{\alpha}}}+\frac{\hat{B}}{D^{\hat{\beta}}}.
$$
The reported fit coefficients are **Dense**: $\alpha=0.61\,(0.57,0.65)$ and $\beta=0.39\,(0.35,0.43)$; **MoLE**: $\alpha=0.56\,(0.49,0.59)$ and $\beta=0.44\,(0.39,0.43)$. The authors conclude that model capacity must keep growing with dataset size, and that MoLE can reduce the number of active parameters needed for the same loss by roughly a factor of $\Delta\approx 2.5\pm 0.2$ for UMA-M [2506.23971].

Training is organized as a two-stage schedule: a direct-force pretraining stage followed by a conservative finetuning stage in which the force head is removed and forces and stresses are computed by autograd to enforce energy conservation. The implementation uses BF16 pretraining, FP32 finetuning, max-atom batching, reduced neighbor counts during pretraining, graph parallelism, fully-sharded data parallelism for large MoLE layers, activation checkpointing, and training runs up to about **10B total parameters**. BF16 alone is reported to degrade accuracy by **20–50%** depending on task [2506.23971].

## 3. Training data, data harmonization, and universal dataset design

The combined dataset used for UMA contains **459,156,663 training examples** and over **30 billion atoms**. The five constituent datasets are **OMat24** with **100,824,585 materials structures**, **OMol25** with **75,889,983 molecular structures**, **OC20++** with **229,054,043 catalysis structures**, **OMC25** with **24,870,226 molecular crystals**, and **ODAC25** with **28,517,826 MOF / DAC structures**. Average structure size ranges from **19 atoms** in OMat24 to **178 atoms** in ODAC25, with a maximum of **350 atoms**, and the combined pairwise element interactions cover nearly the entire periodic table except radioactive elements [2506.23971].

A major obstacle for universal MLIPs is that available datasets are generated under different exchange-correlation functionals, basis sets, core-electron treatments, boundary conditions, and software packages. Total Energy Alignment (TEA) was introduced to address this directly. TEA has two stages: Inner Core Energy Alignment (ICEA) and Atomization Energy Correction (AEC). If two methods have the same atomization energy for a given geometry, total energies can be shifted from one energy scale to another using isolated atom energies; residual fidelity differences are then modeled with a linear scaling
$$
E_{\mathrm{at}}^{[1]}=a\,E_{\mathrm{at}}^{[2]},
$$
which induces the same scaling on forces,
$$
F_i^{[1]}=aF_i^{[2]}.
$$
Using TEA, the authors trained MACE-Osaka24, described as the first open-source neural network potential model based on a unified dataset covering both molecular and crystalline systems [2412.13088].

A different but closely related data philosophy appears in the MAD dataset, which was designed specifically to train universal models that can provide reasonable predictions for arbitrary structures. MAD contains **95,595 structures** and **85 elements** with atomic numbers ranging from **1 to 86, excluding astatine**, and deliberately prioritizes massive atomic diversity over physical plausibility. Starting from stable seed structures, the dataset is expanded through aggressive perturbations such as MC3D-rattled, MC3D-random, MC3D-surface, and MC3D-cluster constructions. All labels are computed with a single consistent DFT protocol based on **Quantum ESPRESSO v7.2**, **AiiDA**, **PBEsol**, **no spin polarization**, **SSSP v1.2 efficiency pseudopotentials**, a **110 Ry** plane-wave cutoff, a **1320 Ry** charge-density cutoff, **Marzari-Vanderbilt-De Vita-Payne cold smearing** with spread **0.01 Ry**, and a **Γ-centered k-point spacing of 0.125 Å⁻¹** in periodic directions. This compact dataset is reported to enable training universal interatomic potentials competitive with models trained on traditional datasets with two to three orders of magnitude more structures [2506.19674].

These two adjacent developments clarify that universality in atomistic ML is not only an architectural question. One route standardizes scale by massive cross-domain training; another aligns heterogeneous corpora almost without redundant calculations; a third deliberately broadens the support of configuration space through aggressive distortion. A plausible implication is that UMA-scale models depend simultaneously on model capacity, dataset breadth, and coherent reference labeling [2506.23971].

## 4. Cross-domain empirical performance

UMA is evaluated as a zero-shot or near-zero-shot universal model across materials, catalysis, molecules, molecular crystals, and MOFs. On **Matbench Discovery**, UMA-M achieves **F1 = 0.930**, reported as the best to date. On **OC20 OOD-both**, the adsorption-energy MAE is **70.2 meV** for UMA-S, **46.5 meV** for UMA-M, and **43.5 meV** for UMA-L, compared with **306.5 meV** for eqV2-OC20 and **343.3 meV** for GemNet-OC20. On **AdsorbML**, success rate rises to **71.12%** for UMA-M and **74.41%** for UMA-L, versus **60.80%** for eqV2-OC20 and **54.88%** for GemNet-OC20. The paper highlights about a **25% improvement in success rate** for UMA-L [2506.23971].

On the molecular domain, the challenging **OMol25** test splits show that UMA improves over a domain-specific eSEN baseline. On the test **OOD-Comp** split, UMA-M reaches **0.42 meV/atom** energy MAE and **3.89 meV/Å** force MAE; UMA-L improves further to **0.34 meV/atom** and **2.92 meV/Å**. In ligand-strain prediction, UMA-M reaches **2.45 meV** MAE, which the authors note is at or below chemical accuracy. The same evaluation suite includes protein-ligand interaction, ionization/affinity, spin-gap, and distance-scaling benchmarks [2506.23971].

On **OMC25** molecular-crystal data, UMA-M obtains **0.8 meV/atom** energy MAE and **3.0 meV/Å** force MAE, while UMA-L reaches **0.6 meV/atom** and **2.3 meV/Å**. On the 7th CCDC CSP blind test polymorph subset, UMA-L achieves **2.49 kJ/mol** lattice-energy MAE and **1.00 match rate**. On the **ODAC** adsorption benchmark for MOFs, UMA-L reaches **291.1 meV** adsorption-energy MAE and **6.5 meV/Å** force MAE, compared with **316.0 meV** and **7.2 meV/Å** for eqV2-ODAC. These results are repeatedly framed as evidence that one model family can now match or exceed many specialized baselines across chemistry and materials tasks [2506.23971].

Related universal models derived from different training strategies show comparable broad-domain behavior. MACE-Osaka24-large, trained on TEA-aligned molecular and crystalline data, attains **0.457 kcal/mol** MAE on **78 drug-like biaryl torsions**, **0.404 eV** MAE on **Transition1x reaction energy**, **0.265 eV** MAE on **Transition1x energy barrier**, and **0.018 Å** overall MAE on bulk-crystal lattice constants. The same study argues that integrating molecular data does not significantly degrade inorganic performance [2412.13088]. This suggests that cross-domain transfer is not exclusive to one architecture, but UMA distinguishes itself by combining that transfer with very large-scale pretraining and a speed-preserving MoLE design.

## 5. Fine-tuning and application workflows

Although UMA is presented primarily as a zero-shot universal model, the literature also treats it as a foundation model that can be adapted to narrower domains. In oxygen-plasma interactions with multilayer WS$_2$, the specific checkpoint used is **`uma-s-1p1`**, the UMA-S variant under the **Open Catalyst 2020 (OC20)** task. The adaptation procedure is an iterative loop: MD exploration with the current model, configuration sampling, DFT labeling, fine-tuning, and evaluation on newly generated configurations. Diversity selection uses **SOAP**, **PCA**, and **FPS**, with SOAP features reduced to **50 principal components** capturing about **99% of the variance**, followed by Farthest Point Sampling. Labels are generated by periodic DFT in **Quantum ESPRESSO** at **PBE + D3 + $U$ + spin**, with **Hubbard $U$ correction of 2.87 eV on W 5d orbitals** [2606.21632].

The training targets in that study are **energy**, **force**, and **stress**, with loss
$$
\mathcal{L}=w_E\frac{1}{B}\sum_{i=1}^B \mathrm{MSE}\!\left(\frac{\hat{E}_i}{N_i},\frac{E_i}{N_i}\right)
+w_F\frac{1}{B}\sum_{i=1}^B \mathrm{Huber}_\delta(\hat{F}_i,F_i)
+w_S\frac{1}{B}\sum_{i=1}^B \mathrm{MSE}(\hat{\sigma}_i,\sigma_i).
$$
The force channel uses Huber loss rather than MSE, while energy and stress use MSE. The production results use layer freezing, with **90/10 train/validation split**, **55 epochs**, **cosine learning-rate schedule**, **5-epoch linear warmup**, and **EMA weights**. Even without fine-tuning, the pretrained model reproduces chemisorbed S and O coverage under **15 eV O$^+$ and O$_2^+$ bombardment**. After iterative fine-tuning, the best reported model, **R3**, reaches **Energy MAE = $4.5\times 10^{-3}$ eV/atom**, **Force MAE = 0.076 eV/Å**, **Stress MAE = 0.034 GPa**, with **$R^2=0.999$** for energy, **$R^2=0.981$** for forces, and **$R^2=0.872$** for stress [2606.21632].

A second application is molecular crystal structure prediction (CSP). In **FastCSP**, UMA is the MLIP core of a fully open, high-throughput workflow that combines random structure generation using **Genarris 3.0** with relaxation and free-energy calculations powered entirely by UMA. The workflow generates random crystal packings across compatible space groups and multiple $Z$ values, compresses them with **Rigid Press**, deduplicates them using **Pymatgen StructureMatcher**, relaxes them with **UMA-Small v1.1**, deduplicates again, and ranks by **0 K lattice energy** or, optionally, by **Helmholtz free energy** or **Gibbs free energy**. The thermodynamic ranking uses
$$
G(T,P)=\min_V\{F(T,V)+PV\},
$$
with the harmonic approximation for $F(T,V)$ and a **Vinet equation of state** in the quasiharmonic approximation [2508.02641].

The benchmark set in FastCSP contains **28 mostly rigid molecules** and **36 crystal structures**. The workflow is reported to have generated the experimental structure for all **28 molecules**, with one caveat: **CILJIQ Form II** required increasing generation from **500 to 1000 structures per space group**. Across the benchmark, **17 of 28** target molecules had an experimental structure or known polymorph ranked as the lattice-energy global minimum, **8 more** were ranked within the **top 4**, all experimental structures were within **5 kJ/mol per molecule** of the global minimum, and the workflow achieved **over 94% recall** within the **top 10** lowest-energy predictions and within **3 kJ/mol** of the global minimum. Comparing UMA-relaxed structures to PBE-D3, the reported metrics are **1.16 kJ/mol** MAE for relaxation energies, **0.94** Spearman rank correlation, **0.22 Å** average RMSD$_{30}$, and **90%** average match rate. Runtime is approximately **15 seconds per relaxation** on an **NVIDIA H100 80 GB GPU**, and a single CSP run can be completed within hours on tens of modern GPUs [2508.02641].

## 6. Uncertainty, reliability, and limitations

A central problem for universal atomistic foundation models is how to determine when predictions are trustworthy without running new DFT calculations. A recent answer is the uncertainty metric $U$ derived from a heterogeneous ensemble of pretrained uMLIPs. The recommended metric is the inverse-RMSE-weighted formulation
$$
U_i^{(1)}=\sqrt{\sum_k w_k\Bigl[\max_j\lVert \mathbf{F}_{i,j,k}-\langle \mathbf{F}_{i,j}\rangle\rVert\Bigr]^2},
$$
with model weights
$$
w_k=\frac{\mathrm{RMSE}_{F,k}^{-1}}{\sum_{k'=1}^K \mathrm{RMSE}_{F,k'}^{-1}}.
$$
The optimal ensemble size is **11 models**. On **OMat24**, the reported correlations are **$\rho=0.82$** for equal-weight $U^{(0)}$, **$\rho=0.87$** for weighted $U^{(1)}$, and **$\rho=0.86$** for weighted-average $U^{(2)}$; the final recommended metric is $U=U^{(1)}$ [2507.21297].

The same study shows that uncertainty can drive filtering and data selection. On OMat24, using $U$, energy RMSE stays below **$0.05\,\mathrm{eV/atom}$** and force RMSE below **$0.06\,\mathrm{eV/\AA}$** up to about **80% coverage**, whereas Orb-confidence reaches comparable accuracy only below roughly **25% coverage** for energy and **40%** for force. A practical threshold
$$
U_c=1\,\mathrm{eV/\AA}
$$
is suggested as a rule of thumb for configurations that can be predicted with RMSE $\le 0.1\,\mathrm{eV/\AA}$, while **$U_c=0.3\,\mathrm{eV/\AA}$** yields about **$0.05\,\mathrm{eV/\AA}$** RMSE. The uncertainty-aware model distillation framework then uses teacher predictions when $U\le U_c$ and DFT labels when $U>U_c$; for tungsten, accuracy comparable to full-DFT training is achieved using only about **4% DFT-labeled configurations**, and for MoNbTaW no additional DFT calculations are required [2507.21297].

Universality, however, is explicitly bounded. In the AML/amons framework, the locality assumption is not exact for strong nonlocal physics such as highly correlated phenomena, Peierls or Jahn–Teller distortions, and metal–insulator transitions; conjugated polymers require larger amons as the query length grows, and proteins or long-range interactions require static or cutoff-based fragments rather than fully relaxed ones [1707.04146]. In TEA-based universal models, the method depends on suitable isolated-atom reference energies, stronger correlation or relativistic systems may be harder to align reliably, a single global scaling factor may be too simple for some cases, and a **4.5 Å** cutoff limits some BCC lattice predictions [2412.13088]. In MAD, consistency is achieved partly by simplifying or neglecting **magnetism, electron correlation, and dispersion**, and convergence falls to about **55%** for the deliberately random **MC3D-random** subset [2506.19674].

The modern UMA family is also constrained by the reference tasks on which it is trained. Because the same structure can have different labels under different DFT settings, UMA includes the DFT task label as an input rather than presuming a single universal energy scale [2506.23971]. Application studies reinforce the same point. In plasma–WS$_2$ chemistry, the pretrained OC20 checkpoint already captures the main production observables, but fine-tuning is motivated precisely because spin polarization and Hubbard $U$ matter for W–S–O chemistry [2606.21632]. In FastCSP, the benchmark is restricted to **mostly rigid molecules**, excludes **$Z'>1$**, cocrystals, salts, hydrates, and solvates, and identifies **Target XVI** as a coverage-dependent failure due to unusual diazide-carbonyl interactions underrepresented in training data [2508.02641]. A common misconception is therefore that “universal” means exact or reference-independent across all chemistry. The literature instead supports a narrower interpretation: universality is practical, data-driven, and strongly conditioned on coverage, labeling conventions, and the target observable.

Source: https://www.emergentmind.com/topics/universal-models-for-atoms-uma