Chatbot Agent 🧠

from typing import TypedDict, List
from langchain_core.messages import HumanMessage
from langchain_ollama import OllamaLLM
from langgraph.graph import StateGraph, START, END
 
class AgentState(TypedDict):
    messages: List[Union[HumanMessage, AIMessage]]
 
ollama_model = "llama3.1:8b"
 
llm = OllamaLLM(temperature=0.8, model=ollama_model)

from IPython.display import Markdown, display
 
def process(state: AgentState) -> AgentState:
    # Extract the content from the last HumanMessage
    prompt = state["messages"][-1].content
    response = llm(prompt)
    state["messages"].append(AIMessage(content=response))
 
    return state
 
def start_node(state: AgentState) -> AgentState:
    prompt = "Hello! Who are you?"
    response = llm(prompt)
    state["messages"].append(AIMessage(content=response))
 
    display(Markdown(f"AI: \n{state['messages'][-1].content}\n"))
    return state
 
def end_node(state: AgentState) -> AgentState:
    state["messages"].append(AIMessage(content="Rate limit of 3 messages reached. Please try again later."))
 
    display(Markdown(f"AI: \n{state['messages'][-1].content}\n"))
 
    return state
 
def take_user_input(state: AgentState) -> AgentState:
    user_input = input("You: ")
 
    state["messages"].append(HumanMessage(content=user_input))
 
    print("HUMAN:", state["messages"][-1].content)
    return state
 
def should_continue(state: AgentState) -> AgentState:
    """Decide whether to continue looping based on the counter."""
 
    display(Markdown(f"AI: \n{state['messages'][-1].content}\n"))
 
 
    if len(state["messages"]) < 6:
        return "loop"
    else:
        return "end"

graph = StateGraph(AgentState)
 
graph.add_node("start_node", start_node)
graph.add_node("take_user_input", take_user_input)
graph.add_node("process", process)
graph.add_node("end_node", end_node)
 
graph.add_edge(START, "start_node")
graph.add_edge("start_node", "take_user_input")
graph.add_edge("take_user_input", "process")
graph.add_conditional_edges(
    "process",
    should_continue,
    {
        # Edge: Node
        "loop": "take_user_input",
        "end": "end_node",
    },
)
graph.add_edge("end_node", END)
 
app = graph.compile()

OUTPUT:
 
AI: "I'm an artificial intelligence model known as Llama. Llama stands for Large Language Model Meta AI."
 
HUMAN: Rasputin
AI:"The infamous Grigori Yefimovich Rasputin! A figure shrouded in mystery and controversy, associated with the downfall of the Russian monarchy. Here's a brief overview:
Early Life:
Born around 1869 in Siberia, Russia, Rasputin was born into a peasant family. His early life is somewhat of a myth, but it's believed he was raised in an Orthodox Christian household and developed a strong  ...."
 
HUMAN: What about the song ?
AI: "I don't think we were discussing a specific song. This conversation just started. Would you like to talk about a particular song or ask for recommendations? I'm here to help!"
 
HUMAN: I forgot you do not have memory XD
AI: "That's a clever reference to the fact that I'm a large language model, I don't have personal memories or experiences like humans do. Each time you interact with me, it's a new conversation and I start from a blank slate. But I'll play along - how can I help you today?"
 
AI: "Rate limit of 3 messages reached. Please try again later."
"

ollama_model = "llama3.1:8b"
 
llm = OllamaLLM(temperature=0.8, model=ollama_model)
 
memory = ConversationSummaryMemory(llm=llm)
memory.clear()
 
conversation = ConversationChain(
    llm=llm,
    memory=memory,
)

ollama_model = "llama3.1:8b"
 
llm = OllamaLLM(temperature=0.8, model=ollama_model)
 
memory = ConversationSummaryMemory(llm=llm)
memory.clear()
 
conversation = ConversationChain(
    llm=llm,
    memory=memory,
)

def process(state: AgentState) -> AgentState:
    # Extract the content from the last HumanMessage
    prompt = state["messages"][-1].content
    response = conversation.predict(input=prompt)
    state["messages"].append(AIMessage(content=response))
    return state
 
def start_node(state: AgentState) -> AgentState:
    prompt = "Hello! Who are you?"
    response = conversation.predict(input=prompt)
    state["messages"].append(AIMessage(content=response))
 
    display(Markdown(f"AI: \n{state['messages'][-1].content}\n"))
    return state

OUTPUT:
 
HUMAN: Hi!
AI: "I'm an artificial intelligence model known as Llama. Llama stands for Large Language Model Meta AI."
 
HUMAN: Rasputin
AI:"The infamous Grigori Yefimovich Rasputin! A figure shrouded in mystery and controversy, associated with the downfall of the Russian monarchy. Here's a brief overview:
Early Life:
Born around 1869 in Siberia, Russia, Rasputin was born into a peasant family. His early life is somewhat of a myth, but it's believed he was raised in an Orthodox Christian household and developed a strong  ...."
 
HUMAN: What about the song ?
AI: "AI: The song! I think I'm starting to get a sense of where this might be going. I've been trained on a vast collection of music-related texts, including lyrics and discographies from various artists. I believe there's a classic rock song called "Rasputin" by Boney M, released in 1978. The song tells the story of Grigori Rasputin, a Russian mystic who gained significant influence over the Romanov dynasty before his eventual assassination in 1916. Is that the context you were thinking of?"
 
AI: "Rate limit of 3 messages reached. Please try again later."
"