license: apache-2.0
language:
- en
tags:
- OneScience
- fluid-dynamics
library_name: pytorch
LiNO
Model and experiment
This package reproduces the one-dimensional viscous Burgers final-state prediction experiment from the paper. LiNO uses eight light-inspired propagation modules with width 128 and efficient scattering. Initial field and spatial coordinates on a 256-point grid are mapped to the final Burgers field.
The package includes model code, config/config.yaml, training, inference and visualization scripts, the recorded results/, and the selected weight/best_model.pth checkpoint.
Data
Download only the needed file from OneScience/fno with modelscope download OneScience/fno data/burgers_data_R10.mat --repo-type dataset --local_dir ./data. The model reads data/data/burgers_data_R10.mat.
Original dataset: https://www.modelscope.cn/datasets/OneScience/fno Download the dataset before running inference where an external dataset is required. The release configuration uses relative paths, not SCNet personal-directory paths.
Installation and usage
hf download OneScience-Group/LiNO --local-dir ./LiNO
cd LiNO
pip install -r requirements.txt
python scripts/inference.py
python scripts/result.py
For a fresh training run, use an allocated GPU/DCU compute node with a compatible PyTorch build, keep a separate copy of the supplied checkpoint, then run python scripts/train.py from the package root. The full budget is specified in config/config.yaml. Cluster-specific Slurm scripts, if present, need adaptation outside the original cluster.
Training and reported result
The run completed 500 epochs with AdamW and a batch size of four. The best checkpoint was selected on 448 validation samples at epoch 495; no independent test split is claimed.
Validation relative L2 was 0.00816638 versus the paper Table 1 value of 0.0109, about 25.1% lower. This is a validation result, not an independent test score. See results/ for the recorded metrics and visualizations. The supplied checkpoint comes from a completed run; another run may differ.
Source and license
- Paper: https://arxiv.org/abs/2606.03262.
- Dataset source: https://www.modelscope.cn/datasets/OneScience/fno.
- This model package uses the license indicated in the card metadata. Dataset redistribution terms, where not separately stated, must not be inferred from the code license.