Closed jaxels20 closed 1 year ago
The current best performing is model with ID 53, it has
Considerations and drawbacks: We are training the model with early stopping, an we experience the loss on the validation set to increase after only 3-5 epochs
Model with all concept types defined in med-rt and edges between them (making it not bipartite): we get two models: Model 109: Validation AUC: 0.8261 Validation Recall: 0.3137 Validation Accuracy: 0.6440 Validation F1: 0.4684 Validation Precision: 0.9243
Model 111: Validation AUC: 0.8382 Validation Recall: 0.6729 Validation Accuracy: 0.7589 Validation F1: 0.7362 Validation Precision: 0.8126
This issue is made to track the progress in the performance of the model.