English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Recent years have seen the advent of molecular simulation datasets that areorders of magnitude larger and more diverse. These new datasets differsubstantially in four aspects of complexity: 1. Chemical diversity (number ofdifferent elements), 2. system size (number of atoms per sample), 3. datasetsize (number of data samples), and 4. domain shift (similarity of the trainingand test set). Despite these large differences, benchmarks on small and narrowdatasets remain the predominant method of demonstrating progress in graphneural networks (GNNs) for molecular simulation, likely due to cheaper trainingcompute requirements. This raises the question -- does GNN progress on smalland narrow datasets translate to these more complex datasets? This workinvestigates this question by first developing the GemNet-OC model based on thelarge Open Catalyst 2020 (OC20) dataset. GemNet-OC outperforms the previousstate-of-the-art on OC20 by 16% while reducing training time by a factor of 10.We then compare the impact of 18 model components and hyperparameter choices onperformance in multiple datasets. We find that the resulting model would bedrastically different depending on the dataset used for making model choices.To isolate the source of this discrepancy we study six subsets of the OC20dataset that individually test each of the above-mentioned four datasetaspects. We find that results on the OC-2M subset correlate well with the fullOC20 dataset while being substantially cheaper to train on. Our findingschallenge the common practice of developing GNNs solely on small datasets, buthighlight ways of achieving fast development cycles and generalizable resultsvia moderately-sized, representative datasets such as OC-2M and efficientmodels such as GemNet-OC. Our code and pretrained model weights areopen-sourced.

GemNet-OC: Developing Graph Neural Networks for Large and Diverse Molecular Simulation Datasets