search

TunnRL / TunnRL_TBM_maintenance

Working repository for the code of the TunnRL TBM project
MIT License
1 stars 0 forks source link

split loss plot in 2 for training/optimization progress #26

Open tfha opened 2 years ago

tfha commented 2 years ago

In the training plot which is generated for every xxx episode in optimization and training the 4 different loss types have very different scales. In order to give information this plot should be splitter in 2 with different scale on the y-axis.

geograz commented 2 years ago

Can we narrow down which agents are affected by this problem? The TD3 is not, because the actor's and the critic's loss have a similar scale. I think the PPO has that issue. Else?