3 ms·
posits still have Near which is a single NaN like value. the problem with floats is that they have -0, -inf, inf and a bunch of NaN values. float16 for example
by adgjlsfhk1 4y ago
posits still have Near which is a single NaN like value. the problem with floats is that they have -0, -inf, inf and a bunch of NaN values. float16 for example wastes over 1000 of the 65000 numbers or can represent on non-finite numbers.