"Simplifying GNN Performance with Low Rank Kernel Models" Implementation

Paper link: https://arxiv.org/abs/2310.05250

Dependencies: torch, optuna, numpy, xarray, torch_geometric, tqdm

For Table 2 results run: python low_rank_sweep.py --masks balanced --graphs none shift --save_name table2_results

MLP2 results can be obtained by running mlp_run.py

Comparison of different evaluation is run using eval_conventions.py