| license: mit | |
| library_name: pytorch | |
| tags: | |
| - graphconv | |
| - cora | |
| - node-classification | |
| - graph-neural-network | |
| # GRAPHCONV Node Classification on Cora | |
| ## Model Details | |
| - **Architecture**: GRAPHCONV | |
| - **Dataset**: Cora | |
| - **Task**: Node Classification | |
| ## Usage | |
| ```python | |
| from src.prediction import Predictor | |
| # Load model | |
| predictor = Predictor.from_checkpoint("best_model.pt") | |
| # Make predictions | |
| result = predictor.predict(graph_data) | |
| print(f"Predictions: {result.predictions}") | |
| print(f"Confidence: {result.confidence}") | |
| ``` | |
| ## Files | |
| - `best_model.pt` - Trained model checkpoint | |
| - `hyperparameters.json` - Training configuration | |
| - `experiment_summary.json` - Final metrics | |
| ## Training | |
| ```bash | |
| python scripts/train.py \ | |
| --dataset cora \ | |
| --model graphconv \ | |
| --epochs 150 \ | |
| --lr_scheduler | |
| ``` | |