JK... All you have to do is take the max number of the bit width in the value (8 bits = 256, 16 bits = 65,535, etc.). Take the full scale range (2g, 4g, etc.) and divide by the max number. That gives you the G value per number in your reading.
Now take your reading and subtract half of the max number, then multiply that by the G per number figure.
You have to remember that the lower numbers are actually going to be negative values.
Sorry for such a clumsy explanation, but hopefully it helps.
Thank you. But is there a manual, I really did not find it online. I am a little confused.
Look up the datasheet for the chip. It's an invensense 6000 something.