mean reward of the 64 samples drawn at each step, averaged over 8 seeds
 step  no baseline  leave-one-out
    1        0.169          0.169
   25        0.511          0.685
   50        0.718          0.881
  100        0.824          0.890
  150        0.842          0.891
  200        0.842          0.890
  300        0.844          0.891
         none: step at which the 10-step average first reaches 0.8: [71, 64, 52, 61, 58, None, None, None]
               per-seed reward over the last 50 steps          : [0.874 0.874 0.875 0.997 0.999 0.625 0.75  0.749]
               seeds finishing above 0.99                      : 2 of 8
leave-one-out: step at which the 10-step average first reaches 0.8: [None, 37, None, None, 31, 49, 29, 31]
               per-seed reward over the last 50 steps          : [0.75  0.875 0.75  0.75  1.    1.    1.    1.   ]
               seeds finishing above 0.99                      : 4 of 8
random-policy reward (1/V) = 0.167
