gnn_training.log

2025-07-18 11:46:15,606 - INFO - Starting GNN training pipeline...
2025-07-18 11:46:15,607 - INFO - Using device: cpu
2025-07-18 11:46:15,645 - INFO - Using Apple Silicon MPS acceleration
2025-07-18 11:46:15,645 - INFO - Loading Cora dataset...
2025-07-18 11:46:25,022 - INFO - Dataset: Cora()
2025-07-18 11:46:25,022 - INFO - Number of graphs: 1
2025-07-18 11:46:25,026 - INFO - Number of features: 1433
2025-07-18 11:46:25,027 - INFO - Number of classes: 7
2025-07-18 11:46:25,027 - INFO - Number of nodes: 2708
2025-07-18 11:46:25,028 - INFO - Number of edges: 10556
2025-07-18 11:46:25,028 - INFO - Average node degree: 3.90
2025-07-18 11:46:25,028 - INFO - Training nodes: 140
2025-07-18 11:46:25,029 - INFO - Validation nodes: 500
2025-07-18 11:46:25,029 - INFO - Test nodes: 1000
2025-07-18 11:46:25,030 - INFO - Data split ratios - Train: 0.052, Val: 0.185, Test: 0.369
2025-07-18 11:46:25,131 - INFO - Creating graph visualization...
2025-07-18 11:49:24,214 - INFO - Starting GNN training pipeline...
2025-07-18 11:49:24,215 - INFO - Using device: cpu
2025-07-18 11:49:24,215 - INFO - Using Apple Silicon MPS acceleration
2025-07-18 11:49:24,215 - INFO - Loading Cora dataset...
2025-07-18 11:49:24,221 - INFO - Dataset: Cora()
2025-07-18 11:49:24,221 - INFO - Number of graphs: 1
2025-07-18 11:49:24,222 - INFO - Number of features: 1433
2025-07-18 11:49:24,223 - INFO - Number of classes: 7
2025-07-18 11:49:24,223 - INFO - Number of nodes: 2708
2025-07-18 11:49:24,224 - INFO - Number of edges: 10556
2025-07-18 11:49:24,224 - INFO - Average node degree: 3.90
2025-07-18 11:49:24,224 - INFO - Training nodes: 140
2025-07-18 11:49:24,225 - INFO - Validation nodes: 500
2025-07-18 11:49:24,225 - INFO - Test nodes: 1000
2025-07-18 11:49:24,225 - INFO - Data split ratios - Train: 0.052, Val: 0.185, Test: 0.369
2025-07-18 11:49:24,244 - INFO - Creating graph visualization...
2025-07-18 11:52:01,661 - INFO - Starting GNN training pipeline...
2025-07-18 11:52:01,661 - INFO - Using device: cpu
2025-07-18 11:52:01,662 - INFO - Using Apple Silicon MPS acceleration
2025-07-18 11:52:01,662 - INFO - Loading Cora dataset...
2025-07-18 11:52:01,669 - INFO - Dataset: Cora()
2025-07-18 11:52:01,670 - INFO - Number of graphs: 1
2025-07-18 11:52:01,672 - INFO - Number of features: 1433
2025-07-18 11:52:01,673 - INFO - Number of classes: 7
2025-07-18 11:52:01,673 - INFO - Number of nodes: 2708
2025-07-18 11:52:01,674 - INFO - Number of edges: 10556
2025-07-18 11:52:01,674 - INFO - Average node degree: 3.90
2025-07-18 11:52:01,674 - INFO - Training nodes: 140
2025-07-18 11:52:01,675 - INFO - Validation nodes: 500
2025-07-18 11:52:01,675 - INFO - Test nodes: 1000
2025-07-18 11:52:01,675 - INFO - Data split ratios - Train: 0.052, Val: 0.185, Test: 0.369
2025-07-18 11:52:01,692 - INFO - Creating graph visualization...
2025-07-18 11:54:14,017 - INFO - Starting GNN training pipeline...
2025-07-18 11:54:14,017 - INFO - Using device: cpu
2025-07-18 11:54:14,018 - INFO - Using Apple Silicon MPS acceleration
2025-07-18 11:54:14,018 - INFO - Loading Cora dataset...
2025-07-18 11:54:14,023 - INFO - Dataset: Cora()
2025-07-18 11:54:14,024 - INFO - Number of graphs: 1
2025-07-18 11:54:14,025 - INFO - Number of features: 1433
2025-07-18 11:54:14,026 - INFO - Number of classes: 7
2025-07-18 11:54:14,026 - INFO - Number of nodes: 2708
2025-07-18 11:54:14,026 - INFO - Number of edges: 10556
2025-07-18 11:54:14,026 - INFO - Average node degree: 3.90
2025-07-18 11:54:14,027 - INFO - Training nodes: 140
2025-07-18 11:54:14,027 - INFO - Validation nodes: 500
2025-07-18 11:54:14,027 - INFO - Test nodes: 1000
2025-07-18 11:54:14,028 - INFO - Data split ratios - Train: 0.052, Val: 0.185, Test: 0.369
2025-07-18 11:54:14,048 - INFO - Creating graph visualization...
2025-07-18 11:54:20,240 - INFO - Graph visualization saved to graph_visualization.png
2025-07-18 11:54:20,240 - INFO - Training configuration: {'hidden_dim': 32, 'num_layers': 2, 'dropout': 0.5, 'learning_rate': 0.001, 'weight_decay': 0.0005, 'epochs': 200, 'patience': 20, 'attention_heads': 8}
2025-07-18 11:54:20,241 - INFO - 
==================================================
2025-07-18 11:54:20,241 - INFO - Training GCN
2025-07-18 11:54:20,241 - INFO - ==================================================
2025-07-18 11:54:20,242 - INFO - Training GCN model...
2025-07-18 11:54:20,441 - INFO - GCN Model initialized with 1433 input features, 32 hidden dim, 7 classes
2025-07-18 11:54:20,442 - INFO - Model parameters: 46,119
2025-07-18 11:54:21,583 - INFO - Epoch 20/200 - Train Loss: 1.9194, Train Acc: 0.7857, Val Loss: 1.9314, Val Acc: 0.6880
2025-07-18 11:54:21,830 - INFO - Epoch 40/200 - Train Loss: 1.8824, Train Acc: 0.8714, Val Loss: 1.9118, Val Acc: 0.7580
2025-07-18 11:54:22,072 - INFO - Epoch 60/200 - Train Loss: 1.8367, Train Acc: 0.9214, Val Loss: 1.8873, Val Acc: 0.7680
2025-07-18 11:54:22,313 - INFO - Epoch 80/200 - Train Loss: 1.7875, Train Acc: 0.8857, Val Loss: 1.8592, Val Acc: 0.7740
2025-07-18 11:54:22,431 - INFO - Early stopping at epoch 90
2025-07-18 11:54:22,443 - INFO - GCN Final Test Accuracy: 0.7930
2025-07-18 11:54:22,445 - INFO - GCN training completed and saved
2025-07-18 11:54:22,445 - INFO - 
==================================================
2025-07-18 11:54:22,445 - INFO - Training GraphSAGE
2025-07-18 11:54:22,445 - INFO - ==================================================
2025-07-18 11:54:22,446 - INFO - Training GraphSAGE model...
2025-07-18 11:54:22,459 - INFO - GraphSAGE Model initialized with 1433 input features, 32 hidden dim, 7 classes
2025-07-18 11:54:22,461 - INFO - Model parameters: 92,199
2025-07-18 11:54:22,972 - INFO - Epoch 20/200 - Train Loss: 1.9226, Train Acc: 0.2929, Val Loss: 1.9494, Val Acc: 0.2020
2025-07-18 11:54:23,306 - INFO - Epoch 40/200 - Train Loss: 1.8653, Train Acc: 0.4214, Val Loss: 1.9226, Val Acc: 0.2440
2025-07-18 11:54:23,640 - INFO - Epoch 60/200 - Train Loss: 1.7637, Train Acc: 0.8000, Val Loss: 1.8790, Val Acc: 0.4360
2025-07-18 11:54:23,976 - INFO - Epoch 80/200 - Train Loss: 1.6420, Train Acc: 0.8857, Val Loss: 1.8200, Val Acc: 0.6080
2025-07-18 11:54:24,300 - INFO - Epoch 100/200 - Train Loss: 1.4988, Train Acc: 0.9500, Val Loss: 1.7470, Val Acc: 0.6640
2025-07-18 11:54:24,630 - INFO - Epoch 120/200 - Train Loss: 1.3248, Train Acc: 0.9571, Val Loss: 1.6629, Val Acc: 0.7080
2025-07-18 11:54:24,989 - INFO - Epoch 140/200 - Train Loss: 1.1383, Train Acc: 0.9714, Val Loss: 1.5735, Val Acc: 0.7480
2025-07-18 11:54:25,308 - INFO - Epoch 160/200 - Train Loss: 1.0036, Train Acc: 0.9786, Val Loss: 1.4778, Val Acc: 0.7640
2025-07-18 11:54:25,632 - INFO - Epoch 180/200 - Train Loss: 0.8511, Train Acc: 0.9714, Val Loss: 1.3865, Val Acc: 0.7800
2025-07-18 11:54:25,739 - INFO - Early stopping at epoch 187
2025-07-18 11:54:25,749 - INFO - GraphSAGE Final Test Accuracy: 0.7680
2025-07-18 11:54:25,751 - INFO - GraphSAGE training completed and saved
2025-07-18 11:54:25,752 - INFO - 
==================================================
2025-07-18 11:54:25,752 - INFO - Training GAT
2025-07-18 11:54:25,753 - INFO - ==================================================
2025-07-18 11:54:25,753 - INFO - Training GAT model...
2025-07-18 11:54:25,776 - INFO - GAT Model initialized with 1433 input features, 32 hidden dim, 7 classes, 8 attention heads
2025-07-18 11:54:25,778 - INFO - Model parameters: 369,429
2025-07-18 11:54:28,023 - INFO - Epoch 20/200 - Train Loss: 1.8903, Train Acc: 0.7857, Val Loss: 1.9061, Val Acc: 0.7900
2025-07-18 11:54:28,464 - INFO - Epoch 40/200 - Train Loss: 1.7788, Train Acc: 0.9143, Val Loss: 1.8423, Val Acc: 0.7960
2025-07-18 11:54:28,605 - INFO - Early stopping at epoch 46
2025-07-18 11:54:28,618 - INFO - GAT Final Test Accuracy: 0.8190
2025-07-18 11:54:28,622 - INFO - GAT training completed and saved
2025-07-18 11:54:28,622 - INFO - Creating training curves...
2025-07-18 11:54:29,106 - INFO - Training curves saved to training_curves.png
2025-07-18 11:54:29,106 - INFO - Creating embeddings visualization...
2025-07-18 11:54:36,376 - INFO - Embeddings visualization saved to embeddings_tsne.png
2025-07-18 11:54:36,377 - INFO - Saving results summary...
2025-07-18 11:54:36,377 - INFO - Results summary saved to results_summary.json
2025-07-18 11:54:36,378 - INFO - 
============================================================
2025-07-18 11:54:36,378 - INFO - FINAL RESULTS COMPARISON
2025-07-18 11:54:36,378 - INFO - ============================================================
2025-07-18 11:54:36,378 - INFO - GCN          - Test Accuracy: 0.7930
2025-07-18 11:54:36,378 - INFO - GraphSAGE    - Test Accuracy: 0.7680
2025-07-18 11:54:36,379 - INFO - GAT          - Test Accuracy: 0.8190
2025-07-18 11:54:36,379 - INFO - 
Best performing model: GAT with accuracy: 0.8190
2025-07-18 11:54:36,379 - INFO - 
All training artifacts saved:
2025-07-18 11:54:36,379 - INFO - - Model checkpoints: best_*_model.pth
2025-07-18 11:54:36,379 - INFO - - Full models: *_full_model.pkl
2025-07-18 11:54:36,380 - INFO - - Training curves: training_curves.png
2025-07-18 11:54:36,380 - INFO - - Embeddings visualization: embeddings_tsne.png
2025-07-18 11:54:36,380 - INFO - - Graph visualization: graph_visualization.png
2025-07-18 11:54:36,380 - INFO - - Results summary: results_summary.json
2025-07-18 11:54:36,380 - INFO - - Training logs: gnn_training.log
2025-07-18 11:54:36,381 - INFO - 
GNN training pipeline completed successfully!