AI Paper Atlas

Research map / Probabilistic and scientific modeling / Molecules and proteins

Variational Monte Carlo on a Budget — Fine-tuning pre-trained Neural Wavefunctions

Michael Scherbela, Leon Gerard, Philipp Grohs

NeurIPS 2023 · Probabilistic and scientific modeling › Molecules and proteins · approach: Machine-learned interatomic potentials · 1 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
use improved architecture on prior dataset (ablation)pre-trained wavefunction · 3 heavy atoms zero-shotenergy error: 76%Improved
use improved architecture on prior dataset (ablation)pre-trained wavefunction · 3 heavy atoms after 4k fine-tuningenergy error: 35%Improved
add normal-mode distortion data augmentation (ablation)pre-trained wavefunction · 3 heavy atoms fine-tunedenergy error: 12%Improved
enlarge training set to 98 molecules with torsional conformers (ablation)pre-trained wavefunction · 3 heavy atoms zero-shotenergy error: 98%Improved
enlarge training set to 98 molecules (ablation)pre-trained wavefunction · 3 heavy atoms fine-tunedenergy errorImproved
increase determinants from 4 to 8 (ablation)pre-trained wavefunction · 3 heavy atoms fine-tunedenergy error: 8%Improved
increase pre-training steps from 128k to 256k (ablation)pre-trained wavefunction · 3 heavy atoms fine-tunedenergy error: 21%Improved
fine-tune 700 steps per geometrypre-trained wavefunction · bicyclobutane dis-TSrelative energy error: 183.8 -> 94.5Improved
fine-tune 4000 steps across geometriespre-trained wavefunction · rotated propadieneenergy variation: below 1.6 mHaImproved
train longer with rotation augmentation (170k to 256k epochs)pre-trained wavefunction · rotated propadieneenergy variation: ±3 -> ±1Improved
fine-tune 4k steps over 10 geometriespre-trained wavefunction · twisted propadienemax deviation along path: 40 -> 2Improved
add pre-trainingneural wavefunction · 3 heavy atoms energy vs stepsenergy error: 1-2 orders lower up to 16k stepsMixed

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