About total_loss = (a_loss + c_losss).mean()

[KGBUSH]

I suppose in the paper, authors maybe calculate the loss for actor and critics respectively. I was wondering it really works better when you do "total_loss = (a_loss + c_losss).mean()" in the loss_func()?

Thanks

[jakkarn]

The sum of gradients is the same as gradients of the sum.