RandomSearchNNs

[NeurIPS 2025] Efficient and Expressive Graph Learning via Random Search Neural Networks

Instructions for Reproducibility

In the generation directory, use the following command to download and preprocess datasets from PyTorch Geometric:
```
source generate_benchmarks.sh
```
Once the datasets are downloaded and preprocessed, in the root directory, use the following command to train models and save results to the results directory.
```
source search_benchmarks_rws.sh
```
After training the models and saving the results, load the results in the load_results.ipynb notebook located in the results directory.