AI Paper Atlas

Research map / Probabilistic and scientific modeling / Molecules and proteins

SaVeNet: A Scalable Vector Network for Enhanced Molecular Representation Learning

Sarp Aykent, Tian Xia

NeurIPS 2023 · Probabilistic and scientific modeling › Molecules and proteins · approach: Diffusion for molecular conformations · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add direction noise (DN) during training (ablation)SAVENET (6 layers, no VA) · QM9std. MAE: 0.79 -> 0.77Improved
scale from 6 to 8 layers (ablation)SAVENET without VA · QM9std. MAE: 0.77 -> 0.71Improved
add vector activations (VA) to 8-layer model (ablation)SAVENET · QM9std. MAE: 0.71 -> 0.69Improved
expand model from 8 to 12 layers (ablation)SAVENET · QM9std. MAE: 0.69 -> 0.63Improved
add vector activations (VA) to 6-layer model, with direction noise (ablation)SAVENET · QM9log MAE: -5.85 -> -5.95Improved
scale from 8 to 12 layers (ablation)SAVENET · QM9log MAE: -6.04 -> -6.14Improved
use SAVENET-B instead of EquiformerSAVENET · QM9 efficiencylatency (ms): 337.3 -> 23.1Improved
use SAVENET-B instead of EquiformerSAVENET · QM9std. MAE: 0.70 -> 0.69Improved
scale SAVENET-B to SAVENET-LSAVENET · QM9std. MAE: 0.69 -> 0.63Improved
scale SAVENET-B to SAVENET-LSAVENET · QM9 efficiencylatency (ms): 23.1 -> 50.6Hurt
scale SAVENET-B to SAVENET-LSAVENET · Molecule3Dstd. MAE: 0.0156 -> 0.0128Improved
SAVENET vs Equiformer on frontier orbital targetsSAVENET-L · QM9epsilon_HOMO MAE (meV): 15.4 -> 16.6Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy