You have a list of 6 column pairs (69 and 70, and so on) that are marked as being "nearly collinear". This means that they are almost exactly the same (except possibly for a scale factor). You need to figure out why these columns in the matrix are almost the same -- are they stimuli with almost identical timing? Or are they motion regressors that are almost identical?
You have a serious problem -- the Matrix inverse average error indicates that your results are not reliable.
Without more information, we cannot diagnose your problem any further.