All Steps: Pretrained Models

A model is trained under a fixed budget on all three training steps in the benchmark and included if training is technically possible under these constraints. The U-Net and LGCNN models are stored in .pt files, the default file type for the PyTorch library, with a configuration file in .yaml format containing the hyperparameters.

The configuration file can be used to automatically instantiate the correct model with our Github repository: github.com/JuliaPelzer/Heat-Plume-Prediction/tree/AllIn1/LGCNN/benchmarkingdata (commit: 83bf879). More details can be found in the implementation. The training code for the other models is available at github.com/kotserge/benchmarking-heat-pump (commit: 1eb5e7a).