You need to experiment with each model's two versions: base vs. instruct-tuned, fine tune both with your Fortran2Cpp dataset and compare performance.
To narrow down the scope of the paper and quickly wrap it up: we could also just only use base versions of all models and generate the complete results.
Table 1 and 2: focus on base model's fine tuning using our dataset
please focus on this first
New Table 3 and 4: focus on instruct model's fine tuning using our dataset
You need to experiment with each model's two versions: base vs. instruct-tuned, fine tune both with your Fortran2Cpp dataset and compare performance.
To narrow down the scope of the paper and quickly wrap it up: we could also just only use base versions of all models and generate the complete results.
Table 1 and 2: focus on base model's fine tuning using our dataset
New Table 3 and 4: focus on instruct model's fine tuning using our dataset