441935 (3) [Avatar] Offline
#1
In the for loop first line is:
choice = np.random.choice(arms,p = av_softmax) # this will work but it will return a float like 0.67823...
counts[choice] += 1 # as a result this will fail

I think the idea there was to use the size of the arms array not the arms array probabilities in here?
choice = np.random.choice(len(arms),p = av_softmax) or replace len(arms) with n

Really enjoying the book!

Thanks
Jed
Brandon B (5) [Avatar] Offline
#2
441935 wrote:In the for loop first line is:
choice = np.random.choice(arms,p = av_softmax) # this will work but it will return a float like 0.67823...
counts[choice] += 1 # as a result this will fail

I think the idea there was to use the size of the arms array not the arms array probabilities in here?
choice = np.random.choice(len(arms),p = av_softmax) or replace len(arms) with n

Really enjoying the book!

Thanks
Jed


Good catch, thanks! We'll make sure that's addressed in the next update.