3 ms·
The measures that drop exponentially like val/bpb and train/loss you should put the x-axis in log-scale. That will better show you if it's converged
by bravura 1y ago
The measures that drop exponentially like val/bpb and train/loss you should put the x-axis in log-scale. That will better show you if it's converged
- sammyd56 1y agoGreat call, thankyou - I switched to log scale for those metrics - agree that it is much clearer.
- bravura 1y agoSorry fat fingers. It should be the y axis that is log scale, not x axis. (Sometimes both is good.) Did you notice the inflection point in which the loss drops faster than expected in the top graph? Maybe you should let it run more…