Skip to content

Adversarr/LRSA-Operator

Repository files navigation

Sample Code for "Simple yet Effective: Low-Rank Spatial Attention for Neural Operators"

arXiv

Download the checkpoints here.

# Setup
uv sync --dev

# Run the preprocessing scripts to get h5 datasets for train/eval.

# Evaluation
# 1. Darcy:  Mean RRMSE: 0.004316
python training/time_independent/infer.py +ckpt_path=checkpoints/darcy-release.ckpt dataset=darcy +num_infer_samples=200 +dryrun=true
# 2. Elasticity: Mean RRMSE: 0.003079
python training/time_independent/infer.py +ckpt_path=checkpoints/elasticity-release.ckpt dataset=elasticity +num_infer_samples=200 +dryrun=true
# 3. Airfoil (Naca): Mean RRMSE: 0.003755
python training/time_independent/infer.py +ckpt_path=checkpoints/naca-release.ckpt dataset=naca dataset.val_file=data/naca/val.h5 +num_infer_samples=200 +dryrun=true
# 4. Navier Stokes: Mean RRMSE: 0.042079
python training/time_dependent_ar/infer.py +ckpt_path=checkpoints/navierstokes-release.ckpt +num_infer_samples=200 +dryrun=true
# 5. Pipe: Mean RRMSE: 0.002306
python training/time_independent/infer.py +ckpt_path=checkpoints/pipe-release.ckpt dataset=pipe +num_infer_samples=200 +dryrun=true
# 6. Plasticity: Mean RRMSE: 0.000465
python training/time_dependent/infer.py +ckpt_path=checkpoints/plasticity-release.ckpt +num_infer_samples=200 +dryrun=true

About

Sample Code for "Simple yet Effective: Low-Rank Spatial Attention for Neural Operators"

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors