1. bash run: conda env create -f environment.yml
2. enter the path: /home/your-username/SOS/inference
3. bash run: export CUDA_VISIBLE_DEVICES=0,1 # if you have 4 GPUs: export CUDA_VISIBLE_DEVICES=0,1,2,3
4. select a Huggingface model name from the file "model_selection.text"
5. bash run: python main.py --hf_model_name slected_hf_model_name

example:

a. python main.py --hf_model_name deepseek-ai/DeepSeek-V2-Lite-Chat

b. python main.py --hf_model_name deepseek-ai/DeepSeek-V2-Lite-Chat --sample_type base

default: --sample_type base\

-- base: Random sampling of agents excluding rare professions\ -- stratified: Stratified sampling of agents excluding rare professions\ -- uniform: Uniform sampling of agents excluding rare professions\

c. python main.py --hf_model_name deepseek-ai/DeepSeek-V2-Lite --sample_type stratified --num_movies 5 --agents_per_movie 15 --rate_num 5 --min_count 100

default:\ --num_movies 5: The number of movies sampled randomly is 100\ --agents_per_movie 15: The number of agents rating each movie is 100\ --rate_num 5: To avoid randomness, each agent rates each movie 3 times\ --min_count 100: Exclude agents whose number in the corresponding domain is less than 50\

note:

1. only run example a now
2. if you are running a model, please add "(running)" behind the slected_hf_model_name in the file "model_selection.text"