user10365714
Reputation: 1
Total model loss is greater than sum of the the model's outputs loss
I'm working on a multi input multi output which take an image and a numpy array as an input with output of 2 regression values , while training i notice that the total loss of the model is greater than the sum of the two outputs , what does it mean ?
Upvotes: 0
Views: 42
Answers (1)
Aashay Sharma
Reputation: 31
I am unable to completely understand your problem, but from what I can comprehend is that unless your val loss and total loss is approximately near by then your model is going good, basically both of your losses should not differ much. For this kind of problems try and use tensorBoard to get graphical output for your model performance and check if model loss and val loss are varying closely or is there any hindrance.
Upvotes: 1
Related Questions
- Loss not Converging for CNN Model
- Loss not decreasing - Pytorch
- Why does the loss decreases and the accuracy dosen't increases? PyTorch
- Loss of Conv-neural-network not decreasing, instead obsoleting
- CNN loss with multiple outputs?
- Validation accuracy increasing but validation loss is also increasing
- Training a model with tensorflow,but loss just won't go down?
- The loss function decreases, but accuracy on train set does not change in tensorflow
- Tensorflow Inception Multiple GPU Training Loss is not Summed?
- Adding more conv layers increases the loss and decreases accuracy, Tensorflow