Multiple GPUs.
Single GPU
Dispatching task to GPU 1 with 81158.0MB free memory. The task is python3 main.py --llm_name llama3.1 --llm_path /home/pengy/llm/llama3.1/70B --tokenizer_path /home/pengy/llm/llama3.1/70B --dtype int8 --parameter 70B --result_folder /home/pengy/projects/8-18-project/Encode_hijick_attack/./llama3.1_attack_results_adv --defense_method 0 --natural 1 --custom 1
