Then have a look at this, regarding positive g's related to the turbulence before. This is from FDR data, 20 seconds around the GA mode activation
The A/P G/A mode recordings have an 1 sample per second interval, so I marked the 1 second time range where the mode must have activated
The vertical accel had 8 samples per second.
The pitch was constant at -1.4 for the last 4 seconds up to 12:38:31.6
(Analysis and Graph Intellectual Property with me)
I'm not saying the CPT behaved perfectly. But the Vertical Acceleration wasn't that clear.