The standard approach for truncating to two decimals is to truncate x * 100
and divide that by 100, which works for numpy arrays:
>>> np.trunc(a * 100) / 100
array([ 0.79000002, 0.79000002, 0.81999999, 0.81999999, 0.79000002], dtype=float32)
Don't be put off by the trailing non-zero digits in the output, those are just an artifact of floating-point imprecision:
>>> np.float32(.79)
0.79000002