* add llama2 native implements * rename configs/eval_llama_7b.py --------- Co-authored-by: zhoufengzhe <zhoufengzhe@pjlab.org.cn>