R2

[gcblqqqqqqqqqqqqqqq]

Hello,The model I trained is RedgCNN, and the dataset used is DrivAerNet_STLs_Combined. The R2 value has consistently been negative and LOSS is decreasing and the performance is decent.

[Mohamedelrefaie]

Could you please share the learning curves (training and val loss over the number of epochs)? We have recently re-trained the model and were able to reproduce the results reported in both the DrivAerNet and DrivAerNet++ papers. For reference, please refer to the main paper for the training hyperparameters used.

[cblgg]

And “ Epoch 1 Training Loss: 0.079081 Time: 4061.08s Average Training R²: -368.4631” “Epoch 1 Validation Loss: 0.0161, Avg Inference Time: 0.3453s Average Validation R²: -58.6274”parameters are: 'exp_name': 'CdPrediction_DrivAerNet_r2_100epochs_5k', 'cuda': True, 'seed': 1, 'num_points': 5000, 'lr': 0.001, 'batch_size': 4, 'epochs': 100, 'dropout': 0.4, 'emb_dims': 512, 'k': 40, 'optimizer': 'adam',