torch squeeze in htl_players causes a crash when testing with only 1 enviornment

[Robokan]

current_action = torch.squeeze(current_action.detach()) around line 91 in hrl_players will cause a crash if you are testing with only on environment. changing it to: current_action = current_action.detach()

fixes the crash.

[xbpeng]

thanks a lot for pointing that out, fixed: https://github.com/nv-tlabs/ASE/commit/f0e41937095014ec972c7f3ae9f3f15a99978a99

[Robokan]

I just found another case in ase_players. Line 88. This one will crash it when testing the llc networks just by itself. current_action = torch.squeeze(current_action.detach()) should be: current_action = current_action.detach()

[xbpeng]

thanks, fixed