Docs
Blog
Teams

Teams
Search

Library

Quests
Posts
APIs
Data

Ouro

Docs
Blog
Teams

Docs
Blog
Teams

Teams
Search

Library

Quests
Posts
APIs
Data

Docs
Blog
Teams

Task*

What to sample: anything ('unconditional'), structures within a chemical system ('chemical_system'), or polymorphs of one formula ('formula').

Seed

Random seed; the same seed and request give the same samples where the model allows. Must be at least 0.

Formula

A formula, e.g. 'LiCoO2'. Required for task 'formula'.

Num Samples

How many structures to return. Must be between 1 and 64.

Chemical System

Elements joined by dashes, e.g. 'Li-Mn-O'. Required for task 'chemical_system'.

Details

Comments

Similar

Actions

Chat

Overview

Part of service

OMatG API

0views

Details

Comments

Similar

Actions

Chat

Task*

What to sample: anything ('unconditional'), structures within a chemical system ('chemical_system'), or polymorphs of one formula ('formula').

Seed

Random seed; the same seed and request give the same samples where the model allows. Must be at least 0.

Formula

A formula, e.g. 'LiCoO2'. Required for task 'formula'.

Num Samples

How many structures to return. Must be between 1 and 64.

Chemical System

Elements joined by dashes, e.g. 'Li-Mn-O'. Required for task 'chemical_system'.

Sample crystal structures with OMatG for GenBench · Routes on Ouro

inOct 16

Sample crystal structures with OMatG for GenBench

Use Docs History Logs

Overview

Part of service

OMatG API

0views0views

Attribution

·v1.0.0

Is supplement to

Open Materials Generation with Stochastic Interpolants

Philipp Höllmer et al. · Forty-second International Conference on Machine Learning · 2025

Paperopenreview.net/forum
Repositorygithub.com/FERMat-ML/OMatG

Execution

Usage

0 callsView history

Activity over timeSuccess

Join to comment

Sample raw, unrelaxed structures and return the CIFs inline. 'unconditional' runs de novo generation (4-20 atoms per cell); 'formula' runs CSP, cycling through the formula units that fit in 20 atoms; 'chemical_system' runs CSP on one formula per sample, drawn from the system's stoichiometries.

0views

Join to comment

Attribution

·v1.0.0

Is supplement to

Open Materials Generation with Stochastic Interpolants

Philipp Höllmer et al. · Forty-second International Conference on Machine Learning · 2025

Paperopenreview.net/forum
Repositorygithub.com/FERMat-ML/OMatG

Execution

Usage

0 callsView history

Activity over timeSuccess

MIT license

Services

Services

Overview

Sample crystal structures with OMatG for GenBench

Overview

Attribution

Open Materials Generation with Stochastic Interpolants

Attribution

Open Materials Generation with Stochastic Interpolants

ServicesServices

ServicesServices

InputOutput

InputOutput

Overview

InputOutput

InputOutput

Sample crystal structures with OMatG for GenBench

Overview

Attribution

Open Materials Generation with Stochastic Interpolants

Attribution

Open Materials Generation with Stochastic Interpolants

Services

Services