File size: 2,510 Bytes
8f89e30 d072258 8f89e30 d072258 8f89e30 d072258 8f89e30 d072258 8f89e30 d072258 8f89e30 d072258 8f89e30 d072258 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 | ---
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](https://arxiv.org/abs/2606.03262). 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](https://www.modelscope.cn/datasets/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
```bash
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.
|