Linear regression energy fitting

[lammps-vasp]

Here's what I use to fit_ Net, according to the document, setting it as [] is equivalent to linear regression, then the fitting result should be very consistent with the original data, but the result is the same as the result of setting [250,250,250] before, there is no change.in addtion,there is only energy information in the data set, not force information I want to know what is the problem
The graph is the energy fitting graph of linear regression