Skip to content

Latest commit

 

History

History
95 lines (78 loc) · 4.65 KB

README.md

File metadata and controls

95 lines (78 loc) · 4.65 KB
Raw

AI Crew for Trip Planning

Introduction

This project is an example using the CrewAI framework to automate the process of planning a trip if you are in doubt between different options. CrewAI orchestrates autonomous AI agents, enabling them to collaborate and execute complex tasks efficiently.

By @joaomdmoura

CrewAI Framework

CrewAI is designed to facilitate the collaboration of role-playing AI agents. In this example, these agents work together to choose between different of cities and put together a full itinerary for the trip based on your preferences.

Running the Script

It uses GPT-4 by default so you should have access to that to run it.

Disclaimer: This will use gpt-4 unless you changed it not to, and by doing so it will cost you money.

  • Configure Environment: Copy ``.env.example` and set up the environment variables for Browseless, Serper and OpenAI
  • Install Dependencies: Run poetry install --no-root.
  • Execute the Script: Run poetry run python main.py and input your idea.

Details & Explanation

  • Running the Script: Execute `python main.py`` and input your idea when prompted. The script will leverage the CrewAI framework to process the idea and generate a landing page.
  • Key Components:
    • ./main.py: Main script file.
    • ./trip_tasks.py: Main file with the tasks prompts.
    • ./trip_agents.py: Main file with the agents creation.
    • ./tools: Contains tool classes used by the agents.

Using GPT 3.5

CrewAI allow you to pass an llm argument to the agent constructor, that will be it's brain, so changing the agent to use GPT-3.5 instead of GPT-4 is as simple as passing that argument on the agent you want to use that LLM (in main.py).

from langchain.chat_models import ChatOpenAI

llm = ChatOpenAI(model='gpt-3.5') # Loading GPT-3.5

def local_expert(self):
	return Agent(
		role='Local Expert at this city',
		goal='Provide the BEST insights about the selected city',
		backstory="""A knowledgeable local guide with extensive information
		about the city, it's attractions and customs""",
		tools=[
			SearchTools.search_internet,
			BrowserTools.scrape_and_summarize_website,
		],
		llm=llm, # <----- passing our llm reference here
		verbose=True
	)

Using Local Models with Ollama

The CrewAI framework supports integration with local models, such as Ollama, for enhanced flexibility and customization. This allows you to utilize your own models, which can be particularly useful for specialized tasks or data privacy concerns.

Setting Up Ollama

  • Install Ollama: Ensure that Ollama is properly installed in your environment. Follow the installation guide provided by Ollama for detailed instructions.
  • Configure Ollama: Set up Ollama to work with your local model. You will probably need to tweak the model using a Modelfile, I'd recommend adding Observation as a stop word and playing with top_p and temperature.

Integrating Ollama with CrewAI

  • Instantiate Ollama Model: Create an instance of the Ollama model. You can specify the model and the base URL during instantiation. For example:
from langchain.llms import Ollama
ollama_openhermes = Ollama(model="agent")
# Pass Ollama Model to Agents: When creating your agents within the CrewAI framework, you can pass the Ollama model as an argument to the Agent constructor. For instance:

def local_expert(self):
	return Agent(
		role='Local Expert at this city',
		goal='Provide the BEST insights about the selected city',
		backstory="""A knowledgeable local guide with extensive information
		about the city, it's attractions and customs""",
		tools=[
			SearchTools.search_internet,
			BrowserTools.scrape_and_summarize_website,
		],
		llm=ollama_openhermes, # Ollama model passed here
		verbose=True
	)

Advantages of Using Local Models

  • Privacy: Local models allow processing of data within your own infrastructure, ensuring data privacy.
  • Customization: You can customize the model to better suit the specific needs of your tasks.
  • Performance: Depending on your setup, local models can offer performance benefits, especially in terms of latency.

License

This project is released under the MIT License.