implement the input tokenizer in Fortran itself, so no need for the encoding script (#1)
download the model.dat directly from some place (#31)
Once done, this will eliminate all Python dependencies for the normal (default) workflow.
Regarding storing the model.dat files, I personally store it outside of this repository, so that I can use git clean -dfx and not worry about accidentally deleting it.
@scivision thanks. What I am hoping to do is:
model.dat
directly from some place (#31)Once done, this will eliminate all Python dependencies for the normal (default) workflow.
Regarding storing the model.dat files, I personally store it outside of this repository, so that I can use
git clean -dfx
and not worry about accidentally deleting it.