Select runs that logged train/f1
to visualize data in this line chart.
to visualize data in this line chart.
MAC Network Weight Decay and Learning Rate
Created on September 14|Last edited on September 15
Comment
Results
Even a small weight decay of 1e-5 kills the performance of the MAC network. We clearly see that the MAC network learns nothing after a while if we inspect the gradients of the MAC cell.
Select runs that logged train/accuracy
to visualize data in this line chart.
to visualize data in this line chart.
Select runs that logged train/loss
to visualize data in this line chart.
to visualize data in this line chart.
Run set
Add a comment