For the dyad-year (and leader-dyad-year) data, there must be some kind of
information loss in order to work within the limited space available to this
package. This package loads a truncated version of the data in which the trade
values are rounded to three decimal points in order to greatly reduce the
disk space for this package. I do not think this to be terribly problematic,
though I admit I do not like it. If this is a problem for your research
question, you may want to consider not using this function for dyad-year or
leader-dyad-year data.
Be mindful that the data are fundamentally state-year or dyad-year and that
extensions to leader-level data should be understood as approximations for
leaders (leader-dyads) in a given state-year (dyad-year).