Comment on Ask HN: Those with success using GPT-4 for programming – what are you doing?parentComments−acqq3yBut 3000003 is exact in fp32, as well as it is 3000002.5, 3000002.75 before it or 3000003.25, 3000003.5, 3000003.75 after it?https://www.h-schmidt.net/FloatConverter/IEEE754.htmlAnd AFAIK the LLMs, lacking the ability to actually calculate, can't check which numbers are such?−siboehm3yHa, you're right. I should've used 30,000,000. The answer it gave regarding how to do "ties to even" was still correct though: "Evenness" is decided by the least significant bit of the mantissa, not be the "evenness" of the decimal representation.
Comments
But 3000003 is exact in fp32, as well as it is 3000002.5, 3000002.75 before it or 3000003.25, 3000003.5, 3000003.75 after it?
https://www.h-schmidt.net/FloatConverter/IEEE754.html
And AFAIK the LLMs, lacking the ability to actually calculate, can't check which numbers are such?
Ha, you're right. I should've used 30,000,000. The answer it gave regarding how to do "ties to even" was still correct though: "Evenness" is decided by the least significant bit of the mantissa, not be the "evenness" of the decimal representation.