davidMis/hybrid-operator-learning-of-wave-scattering-maps-in-high-contrast-media

Hybrid Helmholtz const_back Paper Checkpoints

This model repository contains the trained checkpoint artifacts used by the paper Hybrid operator learning of wave scattering maps in high-contrast media.

Github repo.

The release is filtered for inference and paper reproduction. FNO directories include best_model_metadata.pkl and best_model_state_dict.pt; scOT directories include config.json and model weights. Optimizer state, scheduler state, trainer state, intermediate checkpoint state, logs, W&B files, and caches are not included.

Contents

• —FNO runs: 16
• —scOT runs: 15
• —Staged files: 62
• —Total payload size: 12.5 GB

Usage

hf download dmis09/hybrid-operator-learning-of-wave-scattering-maps-in-high-contrast-media --repo-type model --local-dir outputs/checkpoints/const_back/paper

The downloaded directory preserves the checkpoint-root layout expected by helmholtz_hybrid.evaluation and run_all.py.