Honestly, I don't know why people store floats in anything other than binary. IEEE 754, or bfloat16, if you don't want all that precision is so much more compac

Honestly, I don't know why people store floats in anything other than binary. IEEE 754, or bfloat16, if you don't want all that precision is so much more compact & efficient to parse. The amount of work the computer does to convert text floating point to binary representations, and the potential for there to be subtle differences/bugs in the encoding are just... tremendous.