Demo：
# import Titan-Analysis
import os
current_directory = os.getcwd()
!git clone https://github.com/Zhuang-Zhuang-Liu/Titan-Analysis.git
sys.path.append( current_directory +'/Titan-Analysis')
from utils.utils import show_images_in_directory,folder_clean,manage_guide_json,data_info_put
from agent_zoo.GroupChat import agent_create,titan_analysis,task_load
# llm api
api_zz_deep,url_deep,model_deep = 'sk-xx','https://api.deepseek.com/v1',"deepseek-chat"
llm_config_deep = {"config_list": [{ "model": model_deep,"base_url": url_deep,"api_key": api_zz_deep,"temperature": 1.0 ,"cache_seed":1 }] }
# load guide to desktop for rag
manage_guide_json(guide_zoo_path=current_directory +'/Titan-Analysis/rag_zoo/data_analysis_guide.json',
action='get', guide_name='demo_da_guide', show_guide=False)
# load agent prompt
prompt_path = current_directory+'/Titan-Analysis/agent_zoo/agent_prompts.json'
with open(prompt_path, "r") as file: agent_prompts = json.load(file)
# load data and task
dataset_card_path = current_directory +'/Titan-Analysis/dataset/demo_dataset_card.json'
data_info = data_info_put(dataset_card_path = dataset_card_path,show_data_info=False)
task_info = task_load()
# analysis
titan_analysis(path="coding",
llm_config = llm_config_deep,
agent_prompts = agent_prompts,
guide_path = current_directory + '/Titan-Analysis/virtual_desktop/da_guide.txt', # str
task_info = task_info,
data_info = data_info,
max_round_num = 27 )Copyright and License This project follows the MIT License Agreement