Hello,
I need some advice regarding the visualization from tensorboard I have obtained for a MLP I have done with custom layers.
Now for the interpretation, I understood that in this model, the model is mostly learning by changing the bias of the last dense layer and changing the only one weight value of the two custom layers.
Now for this other model (simple MLP):
The model is learning is once again learning only with the bias of the last dense layer.
Now i was thinking that I can interpret this as the architecture of the model is not suited for the complexity of the task, and I should change the type of layer (as in this case RNN layers, since i am dealing with time series data).
Did I understood correctly or am I wrong ?
Is the first time I am really using tensorboard and not really sure if my interpretation is good.