GENCO power-flow checkpoints on PFΔ (paper): IEEE 118, tasks 1.1, 1.2, 1.3, 2.3, 3.1, 4.1, 4.2, and 4.3, seeds 1, 2, and 3.
Evaluate on graphkit branch genco-paper-repro-pfdelta. That branch produces the paper numbers. These checkpoints were not run on main. That branch does not load the PFΔ configs.
Training, figures, and tables: GENCO §5.1.
Evaluate
Graphkit requires Python 3.10, 3.11, or 3.12.
git clone -b genco-paper-repro-pfdelta https://github.com/gridfm/gridfm-graphkit.git
cd gridfm-graphkit
pip install -e .
pip install huggingface_hub
TORCH_CUDA_VERSION=$(python -c "import torch; print(torch.__version__ + ('+cpu' if torch.version.cuda is None else ''))")
pip install torch-scatter -f https://data.pyg.org/whl/torch-${TORCH_CUDA_VERSION}.html
export MLFLOW_ALLOW_FILE_STORE=true
Run the commands below from that repository root.
hf download gridfm/pfdelta_task4.3 --repo-type dataset --local-dir pfdelta_task4.3
hf download gridfm/genco-pfdelta --include "task_4.3/seed1/**" --local-dir genco-pfdelta
gridfm_graphkit evaluate \
--config scripts/pfdelta/config/HGNS_PF_pfdelta_bs64_seed1.yaml \
--model_path genco-pfdelta/task_4.3/seed1/last.pt \
--data_path pfdelta_task4.3/case118
For another task or seed, download that dataset and task_<id>/seed<seed>/**. Point --data_path at that dataset's case118 folder and --model_path at genco-pfdelta/task_<id>/seed<seed>/last.pt.
Task 3.1 was not trained separately. Use task_3.1/seed<seed>/best_model_state_dict.epoch_299.pt (the task 1.3 best-validation snapshot) and --data_path pointing at the task 3.1 case118 folder.