Tutorials¶
Step-by-step guides for common workflows, from beginner to advanced.
Beginner¶
| Tutorial | Description |
|---|---|
| Training Your First Model | Load case14 data, configure training, launch with torchrun |
| Evaluation | Load checkpoints, run inference, check constraint violations |
Advanced¶
| Tutorial | Description |
|---|---|
| Multi-Case Training | Train across multiple grid topologies with DDP |
| Custom Models | Add new GNN backends, register in ModelFactory |
| HPC Training | Polaris, Perlmutter setup and job scripts |
Notebooks¶
Runnable Jupyter notebooks walking through end-to-end workflows on case14.
| Notebook | Description |
|---|---|
| Quickstart (case14) | Smallest end-to-end run: load, build, train, plot |
| Dataset exploration | Inspect node/edge types, schemas, target distributions |
| Evaluation walkthrough | Load a checkpoint and measure constraint violations |
| Physics-informed loss | MSE vs. MSE + power-balance penalty, side by side |
| DDP on local GPUs | Multi-GPU training on a single workstation via torchrun |
| Polaris (HPC) | PBS job script and rendezvous setup for ALCF Polaris |