Why do geom_smooth nls and the stand alone nls give different fit results?

Question

When I fit my data using geom_smooth nls, I get a really nice fit. But if I use the stand alone nls function to fit my data using the same equation and starting values, I get a