Skip to content

Latest commit

 

History

History
105 lines (84 loc) · 3.66 KB

agent.md

File metadata and controls

105 lines (84 loc) · 3.66 KB
Raw

React Agent

Agent aims to interact with a language model and use tools to perform actions. It is basically based on the React prompt.

Use Case

import httpx
from loguru import logger
from miniagent.llm.gpt import ChatGPT
from miniagent.llm.base import LLMInput
from miniagent.tools import *
from miniagent.agent import Agent

# define the LLM
api_key="sk-xxx"
llm = ChatGPT(api_key=api_key)

# define tools
tools = ToolList([EmailTool(user_email='miniagent.ai@gmail.com', password='xxx', recipient="rayyang0116@gmail.com"),
                    ArxivTool(),])

# define_agent
agent = Agent(llm=llm, tools=tools)

### Search the related topic from arxiv, and send the results of the search to someone.
### you can change the email rayyang0116.gmail.com, r-yang20@tsinghua.org.cn to you own email
user_input = "Please help me search the realted paper about language agent, and send it to rayyang0116.gmail.com, r-yang20@tsinghua.org.cn."
# start!
agent.execute(input=user_input)

Response Agent

ResponseAgent is designed to interact with a language model (LLM) to generate summaries or answers based on tool outputs. The recommended prompt format is:

PromptTemplate(
    prompt="Content: {action_output} your_prompt",
    input_variables=["action_output", "your_variables"]
)

Use Case

import httpx
from miniagent.agent import ResponseAgent
from miniagent.llm.gpt import ChatGPT
from miniagent.tools import ScrapTool

# define the LLM
api_key="sk-xxx"
llm = ChatGPT(api_key=api_key)
tool = ScrapTool()
scrap_agent = ResponseAgent(
    llm=llm,
    tools=tool,
    prompt=PrompeTemplateV2(prompt="Please summary the content of this website.")
)
output = scrap_agent.execute(url="http://www.maxlikelihood.cn")
print(output)
Clik to check the output of this example.

Likelihood Lab Overview:

Likelihood Lab is a public AI research initiative founded by Maxwell (Mingwen) Liu and a team of experts from various prestigious institutions, including MIT and Stanford. The lab focuses on advancing technology in FinTech, Energy, and Robotics.

Research Areas:

  • The lab conducts research in multiple fields, primarily:
    • FinTech: Topics include reinforcement learning for market making, asset allocation, stock selection using high-frequency data, and financial sentiment analysis.
    • Energy and Robotics: Specific projects in these areas weren't detailed but are part of the lab's scope.

Highlighted Projects:

  • Reinforcement learning in market making
  • Ensemble learning for detecting short-selling
  • Machine learning frameworks for stock selection
  • Generative adversarial networks for day trading chart generation
  • AI applications in games like Mahjong and Gomoku

Events and Conferences:

  • The lab organizes and participates in various conferences and seminars, focusing on machine learning and quantitative finance, including seasonal research conferences and special events like the Asian Quantitative Finance Conference.

Team: The team includes co-founders and key researchers who bring diverse expertise from notable academic and research institutions.

Contact Information:

The lab aims to leverage AI for societal benefits and innovation across its research areas.

Human Agent

HumanAgent serves as an interface for human interaction. It captures input from the user, formats it, and returns it for further processing.

Use Case

from miniagent.agent import HumanAgent
human = HumanAgent()
output = human.execute()