I don't know much about the offsets or hex but I read the chip with your offsets and the ones listed above and they both look the same in the hex editor.
We're using different tools to read the chip and that may explain why the Moates article recommends different addresses. You should learn some basics about using hex and working with addresses and a hex editor. There are times when you might want to hand edit a calibration or when you might want to confirm that a file isn't corrupt that would require the hex edit tool.

The only ones that through a loop into that (or make you think) are the ones that don't have anything at the beginning of the chip and start an 1/8 or so into the hex editor addressing.
Some of the 2.3 cals have data all over the place. I thought my disassembler was broken when I first started working on one because it just wasn't arranged like a normal GM cal. And the coding is different, too. I think the code was generated by a newer compiler than what's used for many of the more common cals.