This is where the params.json and tokenizer.model required to run llama2 are stored.