ADL5519 evaluation board glitch question

I have an ADL5519 evaluation board, and I noticed that, if I feed a low level signal (say 1GHz frequency, -120dBm up to -60dBm) into INHA (or terminate INHA in 50 ohms) I can see a recurring output glitch on OUTA (see scope captures below).  As I increase the input power, the amplitude of the glitches decreases, but they do not disappear.

I am trying to use the detector to recover low frequency pulse train patterns, and this glitch is in the passband of the data I am trying to extract.