From 73c63da4bd98a5906dd13a2eafd5573d3aef6ffc Mon Sep 17 00:00:00 2001 From: ShubhamSaboo Date: Wed, 5 Jun 2024 22:02:05 -0500 Subject: [PATCH] Added multi-agent demo with llama-3 --- .../research_agent_llama3.py | 39 +++++++++++++++++++ 1 file changed, 39 insertions(+) create mode 100644 multi_agent_researcher/research_agent_llama3.py diff --git a/multi_agent_researcher/research_agent_llama3.py b/multi_agent_researcher/research_agent_llama3.py new file mode 100644 index 0000000..778093f --- /dev/null +++ b/multi_agent_researcher/research_agent_llama3.py @@ -0,0 +1,39 @@ +# Import the required libraries +import streamlit as st +from phi.assistant import Assistant +from phi.tools.hackernews import HackerNews +from phi.llm.ollama import Ollama + +# Set up the Streamlit app +st.title("Multi-Agent AI Researcher using Llama-3 🔍🤖") +st.caption("This app allows you to research top stories and users on HackerNews and write blogs, reports and social posts.") + + +# Create instances of the Assistant +story_researcher = Assistant( + name="HackerNews Story Researcher", + role="Researches hackernews stories and users.", + tools=[HackerNews()], + llm=Ollama(model="llama3:instruct", max_tokens=1024) +) + +user_researcher = Assistant( + name="HackerNews User Researcher", + role="Reads articles from URLs.", + tools=[HackerNews()], + llm=Ollama(model="llama3:instruct", max_tokens=1024) +) + +hn_assistant = Assistant( + name="Hackernews Team", + team=[story_researcher, user_researcher], + llm=Ollama(model="llama3:instruct", max_tokens=1024) +) + +# Input field for the report query +query = st.text_input("Enter your report query") + +if query: + # Get the response from the assistant + response = hn_assistant.run(query, stream=False) + st.write(response) \ No newline at end of file