-
Option 1. Build from source
To install using this method,you need to change 'from agents.XXX import XXX' to 'from XXX import XXX' in either run.py or run_gradio.py.git clone https://github.com/aiwaves-cn/agents.git cd agents pip install -e .
-
Option 2. Install via PyPI
It is recommended to install in this way, and every code update will require reinstallation afterwardspip install ai-agents
{ // e.g. for shopping assistant（due to change the search engine，it maybe not work now）
MIN_CATEGORY_SIM = "0.7" ##Threshold for category matching
TOY_INFO_PATH = "[\"your_path1\",\"your_path2_\"......]" #Path to the product database
FETSIZE = "5" #Number of recommended products at a time
SHOPPING_SEARCH = "Your url" # Your search engine
#for all agents
API_KEY = #Your API KEY
PROXY = #Your proxy
API_BASE = # Your api_base
MAX_CHAT_HISTORY = "8" #Longest History
User_Names = "[\"{user_name}\"]" # the name of agents which you want to play
}
Note that if you want to use WebSearchComponent, you also need set the config!
"WebSearchComponent": {
"engine_name": "bing",
"api": {
"google": {
"cse_id": "Your cse_id",
"api_key": "Your api_key"
},
"bing": "Your bing key"
}
}
-
Option 1. Run on your terminal
cd examples python run.py --agent Single_Agent/{target_agent}/config.json -
Option 2. Run on gradio
cd examples # `run_gradio.py` depends on `gradio_backend.py`. python Single_Agent/run_gradio.py --agent Single_Agent/{target_agent}/config.json
-
Option 3. Run on fast api
cd examples python run_backend.py --agent Single_Agent/{target_agent}/config.json
-
Option 1. Run on your terminal
Modify the config.json
For example, if you want to play the agent named "Mike" and "Mary"(Support all Agent)
{ #for all agents API_KEY = #Your API KEY PROXY = #Your proxy MAX_CHAT_HISTORY = "8" #Longest History User_Names = "[\"Mike\",\"Mary\"]" # the name of agents which you want to play }cd examples python run.py --agent Multi_Agent/{target_agent}/config.json -
Option 2. Run on gradio
cd examples/Multi_Agent/{target_agent} # `run_gradio.py` depends on `gradio_backend.py`. python run_gradio.py
Choose the agent you want to perform in the gradio UI.
Note: You need to set environment variables first, if you want to run Fiction-Studio demo on gradio.
export PROXY="xxx" export API_KEY="sk-xxxxxxx"And then
cd examples/Multi_Agent/novel python run_gradio.py -
Option 3. Run on fast api
cd examples python run_backend.py --agent Multi_Agent/{target_agent}/config.json
1.Write your config.json according to template.py(more details refer to the doc)
-
Option 1. Run on your terminal
cd examples python run.py --agent config.json -
Option 2. Run on gradio
cd examples # `run_gradio.py` depends on `gradio_backend.py`. python run_gradio.py --agent your_config.json
-
Option 3. Run on fast api
cd examples python run_backend.py --agent config.json
refer src/agents/LLM/base_LLM.py
def init_LLM(default_log_path,**kwargs):
LLM_type = kwargs["LLM_type"] if "LLM_type" in kwargs else "OpenAI"
log_path = kwargs["log_path"].replace("/",os.sep) if "log_path" in kwargs else default_log_path
if LLM_type == "OpenAI":
LLM = (
OpenAILLM(**kwargs["LLM"])
if "LLM" in kwargs
else OpenAILLM(model = "gpt-3.5-turbo-16k-0613",temperature=0.3,log_path=log_path)
)
return LLMYou can change this method to deploy your own LLM. For example:
def init_LLM(default_log_path,**kwargs):
LLM = Your_LLM()
return LLMAlso, ensure that your LLM's input and output parameters remain consistent with the original LLM, and keep the method name as 'get_response'. For example:
class Your_LLM(LLM):
def __init__(self,**kwargs) -> None:
super().__init__()
def get_response(self,
chat_history,
system_prompt,
last_prompt=None,
stream=False,
functions=None,
function_call="auto",
WAIT_TIME=20,
**kwargs):
return chatglm(**kwargs)Please note that it is essential to ensure consistency in streaming output. For instance, when 'stream=True' is set, the function should return a generator, and when 'stream=False,' it should return a string.
- Option 1. Single agent🤖️
API_KEY = "API_KEY" # change to your api-key,delete it if you do not need it
PROXY = "PROXY" # change to your proxy,delete it if you do not need it
API_BASE = "API_BASE" # change to your api_base,delete it if you do not need it
target = """a shopping assistant help customer to buy the commodity""" # change to your target
os.environ["API_KEY"] = API_KEY
os.environ["PROXY"] = PROXY
sop["config"]["API_KEY"] = API_KEY
sop["config"]["PROXY"] = PROXY cd examples/sop_generation/gen_single_agent
python run.py- Option 2. Multi agent🤖️🤖️🤖️
API_KEY = "API_KEY" # change to your api-key,delete it if you do not need it
PROXY = "PROXY" # change to your proxy,delete it if you do not need it
API_BASE = "API_BASE" # change to your api_base,delete it if you do not need it
target = """a shopping assistant help customer to buy the commodity""" # change to your target
need_coder = True # True if the scene need output code
os.environ["API_KEY"] = API_KEY
os.environ["PROXY"] = PROXY
sop["config"]["API_KEY"] = API_KEY
sop["config"]["PROXY"] = PROXY cd examples/sop_generation/gen_multi_agent
python run.py[x] Option 3. SOP Generation for Multi-Agent Multi agent🤖️🤖️🤖️
[x] Option 4. SOP Generation for Multi-Agent Single agent🤖️🤖️🤖️
For option 3 and option 4, we recommend strongly that you can create your OWN DEMO by clicking the three dots at the top right and selecting Duplicate this Space on Huggingface Space.
We provide an AgentHub, where you can search for interesting Agents shared by us or other developers, try them out or use them as the starting point to customize your own agent. We encourage you to share your customized agents to help others build their own agents more easily! You can share your customized agents by submitting PRs that adds configs and customized codes here. You can also send us your own config files and codes for customized agents by email, and we will share your examples and acknowledge your contribution in future updates!