A question about R_batch[-1,0]

[lwz339]

Hi,bro!
I am sorry to bother you again.
For the terminal state,you use two different versions:
if terminal: # in this case, the terminal reward will be assigned as r_batch[-1] R_batch[-1] = r_batch[-1] # terminal state
and
if terminal: R_batch[-1, 0] = 0 # terminal state.
I think the first one in line with my understanding.Why the R of terminal state equal to zero in your tf version and vanilla-Pensieve?
Thank your very much!