re-run compilation and execution tests of StarCoder for Table 3

[chunhualiao]

the numbers I got for Codellama are different from what I saw in Table 3.

This leads to the need to re-produce the numbers for StarCoder also.

Fortran2Cpp/data/StarCoder2/readable_format_HPEC/cpp]make -k -j all

(.venv) [liaoch@login15:~/workspace-cfs/Fortran2Cpp/data/StarCoder2/readable_format_HPEC/cpp]ls *.o|wc
    204     204    2040
(.venv) [liaoch@login15:~/workspace-cfs/Fortran2Cpp/data/StarCoder2/readable_format_HPEC/cpp]ls *.err|wc
     92      92    1104


make -f makefile-exe-test -k -j execute

(.venv) [liaoch@login15:~/workspace-cfs/Fortran2Cpp/data/StarCoder2/readable_format_HPEC/cpp]ls *.pass |wc
    155     155    2015

[chunhualiao]

The paper mentions StartCoder. However, the git repo has Fortran2Cpp/data/StarCoder2 .

Which model did you fine-tune: StartCoder or StarCoder2 ? @bin123apple

[chunhualiao]

the two numbers (compilation and execution metrics) are slightly different. I updated Table 3 to use mine.