Correct llama3-8b-instruct model file
This must be a mistake because the config tries to use a model file that is different from the one actually being downloaded.
I assumed the downloaded file is what should be used so I corrected the specified model file to that
Signed-off-by: Aleksandr Oleinikov <10602045+tannisroot@users.noreply.github.com>