Add ai_tic_tac_toe_agent folder
This commit is contained in:
Madhu
parent
abb2cffa62
commit
aba8da2c78
3 changed files with 345 additions and 0 deletions
68
ai_agent_tutorials/ai_tic_tac_toe_agent/README.md
Normal file
68
ai_agent_tutorials/ai_tic_tac_toe_agent/README.md
Normal file
|
|
@ -0,0 +1,68 @@
|
|||
# AI vs AI Tic-Tac-Toe Game 🎮
|
||||
|
||||
An interactive Tic-Tac-Toe game where two AI agents powered by different language models compete against each other built on phidata Agent Framework and Streamlit as UI. Watch as GPT-4O battles against either DeepSeek V3 or Google's Gemini 1.5 Flash in this classic game!
|
||||
|
||||
## Features 🌟
|
||||
|
||||
- Player X: Powered by OpenAI's GPT-4o (o1 models as api is yet to roll out for tier)
|
||||
- Player O: Choice between DeepSeek AI or Google's Gemini
|
||||
- Interactive game board using Streamlit
|
||||
- Real-time move visualization
|
||||
- AI Judge to determine game outcomes
|
||||
- Clean and intuitive user interface
|
||||
|
||||
## Prerequisites 📋
|
||||
|
||||
- Python 3.8+
|
||||
- OpenAI API key
|
||||
- Either DeepSeek API key or Google API key (for Player O)
|
||||
|
||||
## How to Run 🚀
|
||||
|
||||
1. **Setup Environment**
|
||||
```bash
|
||||
# Clone the repository
|
||||
git clone https://github.com/Shubhamsaboo/awesome-llm-apps.git
|
||||
cd ai_agent_tutorials/ai_tic_tac_toe_agent
|
||||
|
||||
# Install dependencies
|
||||
pip install -r requirements.txt
|
||||
```
|
||||
|
||||
2. **Configure API Keys**
|
||||
- Get OpenAI API key from [OpenAI Platform](https://platform.openai.com)
|
||||
- Get Google API key from [Google AI Studio](https://aistudio.google.com) (if using Gemini)
|
||||
- Get DeepSeek API key from DeepSeek platform [Deepseek platform](https://www.deepseek.com) (if using DeepSeek v3 model)
|
||||
|
||||
4. **Run the Application**
|
||||
```bash
|
||||
streamlit run ai_tic_tac_toe_agent.py
|
||||
```
|
||||
|
||||
5. **Using the Interface**
|
||||
- Enter your API keys in the sidebar
|
||||
- Click "Start Game" to begin
|
||||
- Watch as the AI agents battle it out!
|
||||
- Monitor the game progress and final results
|
||||
|
||||
## Game Components 🎯
|
||||
|
||||
- **Game Board**
|
||||
- 3x3 interactive grid
|
||||
- Real-time move visualization
|
||||
- Clear symbol placement (X/O)
|
||||
|
||||
- **AI Players**
|
||||
- Player X (GPT-4o): Strategic offensive moves
|
||||
- Player O (DeepSeek/Gemini): Defensive countermoves since this agent started later
|
||||
- AI Judge: Game outcome validation
|
||||
|
||||
- **Game Flow**
|
||||
- Alternating turns between AIs
|
||||
- Move validation and error handling
|
||||
- Winner determination
|
||||
- Draw detection
|
||||
|
||||
## Disclaimer ⚠️
|
||||
|
||||
This is a demonstration project showcasing AI capabilities in game playing. API costs will apply based on your usage of the OpenAI, DeepSeek, or Google APIs.
|
||||
273
ai_agent_tutorials/ai_tic_tac_toe_agent/ai_tic_tac_toe_agent.py
Normal file
273
ai_agent_tutorials/ai_tic_tac_toe_agent/ai_tic_tac_toe_agent.py
Normal file
|
|
@ -0,0 +1,273 @@
|
|||
import re
|
||||
import streamlit as st
|
||||
from phi.agent import Agent
|
||||
from phi.model.openai import OpenAIChat
|
||||
from phi.model.deepseek import DeepSeekChat
|
||||
from phi.model.google import Gemini
|
||||
|
||||
# Streamlit App Title
|
||||
st.title("🤖 Tic-Tac-Toe: AI vs AI")
|
||||
|
||||
# Enhanced Welcome Message
|
||||
with st.chat_message("assistant"):
|
||||
st.markdown("""
|
||||
**Welcome to the Tic-Tac-Toe AI Battle!** 🎮
|
||||
This project pits two advanced AI agents against each other in a classic game of Tic-Tac-Toe.
|
||||
Here's what you need to know:
|
||||
""")
|
||||
st.info("""
|
||||
- **Player X**: Powered by OpenAI's GPT-4.
|
||||
- **Player O**: Powered by DeepSeek's AI or Google's Gemini (if DeepSeek API key is not provided).
|
||||
- **How to Play**: Enter your API keys in the sidebar, click **Start Game**, and watch the AI battle it out!
|
||||
- **Goal**: The first player to get three of their symbols in a row (horizontally, vertically, or diagonally) wins.
|
||||
- **Draw**: If the board fills up without a winner, the game ends in a draw.
|
||||
""")
|
||||
st.markdown("Ready to see who wins? Click the **Start Game** button below! 🚀")
|
||||
|
||||
# Initialize the game board in session state
|
||||
if 'board' not in st.session_state:
|
||||
st.session_state.board = [[None, None, None],
|
||||
[None, None, None],
|
||||
[None, None, None]]
|
||||
|
||||
# Function to display the board with better styling
|
||||
def display_board(board):
|
||||
st.markdown("""
|
||||
<style>
|
||||
.board {
|
||||
display: grid;
|
||||
grid-template-columns: repeat(3, 50px);
|
||||
grid-template-rows: repeat(3, 50px);
|
||||
gap: 5px;
|
||||
margin-bottom: 20px;
|
||||
}
|
||||
.cell {
|
||||
display: flex;
|
||||
align-items: center;
|
||||
justify-content: center;
|
||||
border: 1px solid #ccc;
|
||||
font-size: 20px;
|
||||
font-weight: bold;
|
||||
}
|
||||
</style>
|
||||
""", unsafe_allow_html=True)
|
||||
|
||||
st.write("Current Board:")
|
||||
board_html = '<div class="board">'
|
||||
for row in board:
|
||||
for cell in row:
|
||||
cell_value = cell if cell is not None else " "
|
||||
board_html += f'<div class="cell">{cell_value}</div>'
|
||||
board_html += '</div>'
|
||||
st.markdown(board_html, unsafe_allow_html=True)
|
||||
|
||||
# Function to get the board state as a string
|
||||
def get_board_state(board):
|
||||
rows = []
|
||||
for i, row in enumerate(board):
|
||||
row_str = " | ".join([f"({i},{j}) {cell or ' '}" for j, cell in enumerate(row)])
|
||||
rows.append(f"Row {i}: {row_str}")
|
||||
return "\n".join(rows)
|
||||
|
||||
# Function to check for a winner
|
||||
def check_winner(board):
|
||||
# Check rows
|
||||
for row in board:
|
||||
if row[0] == row[1] == row[2] and row[0] is not None:
|
||||
return row[0]
|
||||
# Check columns
|
||||
for col in range(3):
|
||||
if board[0][col] == board[1][col] == board[2][col] and board[0][col] is not None:
|
||||
return board[0][col]
|
||||
# Check diagonals
|
||||
if board[0][0] == board[1][1] == board[2][2] and board[0][0] is not None:
|
||||
return board[0][0]
|
||||
if board[0][2] == board[1][1] == board[2][0] and board[0][2] is not None:
|
||||
return board[0][2]
|
||||
# Check for draw
|
||||
if all(cell is not None for row in board for cell in row):
|
||||
return "Draw"
|
||||
return None
|
||||
|
||||
# Sidebar for API keys
|
||||
st.sidebar.header("API Keys")
|
||||
openai_api_key = st.sidebar.text_input("Enter your OpenAI API Key", type="password")
|
||||
deepseek_api_key = st.sidebar.text_input("Enter your DeepSeek API Key", type="password")
|
||||
google_api_key = st.sidebar.text_input("Enter your Google API Key (for Gemini)", type="password")
|
||||
|
||||
if openai_api_key:
|
||||
st.session_state.openai_api_key = openai_api_key
|
||||
if deepseek_api_key:
|
||||
st.session_state.deepseek_api_key = deepseek_api_key
|
||||
if google_api_key:
|
||||
st.session_state.google_api_key = google_api_key
|
||||
|
||||
# Initialize agents if API keys are provided
|
||||
if 'openai_api_key' in st.session_state:
|
||||
player_x = Agent(
|
||||
name="Player X",
|
||||
model=OpenAIChat(id="gpt-4o", temperature=0.1, api_key=st.session_state.openai_api_key),
|
||||
instructions=[
|
||||
"You are a Tic-Tac-Toe player using the symbol 'X'.",
|
||||
"Your opponent is using the symbol 'O'. Block their potential winning moves.",
|
||||
"Make your move in the format 'row, col' based on the current board state.",
|
||||
"Strategize to win by placing your symbol in a way that blocks your opponent from forming a straight line.",
|
||||
"Do not include any explanations or extra text. Only provide the move.",
|
||||
"Row and column indices start from 0.",
|
||||
],
|
||||
markdown=True,
|
||||
)
|
||||
|
||||
# Initialize Player O with DeepSeek or Google Gemini
|
||||
if 'deepseek_api_key' in st.session_state:
|
||||
player_o = Agent(
|
||||
name="Player O",
|
||||
model=DeepSeekChat(api_key=st.session_state.deepseek_api_key),
|
||||
instructions=[
|
||||
"You are a Tic-Tac-Toe player using the symbol 'O'.",
|
||||
"Your opponent is using the symbol 'X'. Block their potential winning moves.",
|
||||
"Make your move in the format 'row, col' based on the current board state.",
|
||||
"Strategize to win by placing your symbol in a way that blocks your opponent from forming a straight line.",
|
||||
"Do not include any explanations or extra text. Only provide the move.",
|
||||
"Row and column indices start from 0.",
|
||||
],
|
||||
markdown=True,
|
||||
)
|
||||
elif 'google_api_key' in st.session_state:
|
||||
player_o = Agent(
|
||||
name="Player O",
|
||||
model=Gemini(id="gemini-1.5-flash", api_key=st.session_state.google_api_key),
|
||||
instructions=[
|
||||
"You are a Tic-Tac-Toe player using the symbol 'O'.",
|
||||
"Your opponent is using the symbol 'X'. Block their potential winning moves.",
|
||||
"Make your move in the format 'row, col' based on the current board state.",
|
||||
"Strategize to win by placing your symbol in a way that blocks your opponent from forming a straight line.",
|
||||
"Do not include any explanations or extra text. Only provide the move.",
|
||||
"Row and column indices start from 0.",
|
||||
],
|
||||
markdown=True,
|
||||
)
|
||||
else:
|
||||
st.warning("Please provide either a DeepSeek API key or a Google API key for Player O.")
|
||||
|
||||
judge = Agent(
|
||||
name="Judge",
|
||||
model=OpenAIChat(id="gpt-4o", temperature=0.1, api_key=st.session_state.openai_api_key),
|
||||
instructions=[
|
||||
"You are the judge of a Tic-Tac-Toe game.",
|
||||
"The board is presented as rows with positions separated by '|'.",
|
||||
"Rows are labeled from 0 to 2, and columns from 0 to 2.",
|
||||
"Example board state:",
|
||||
"Row 0: (0,0) X | (0,1) | (0,2) O",
|
||||
"Row 1: (1,0) O | (1,1) X | (1,2) X",
|
||||
"Row 2: (2,0) X | (2,1) O | (2,2) O",
|
||||
"Determine the winner based on this board state.",
|
||||
"The winner is the player with three of their symbols in a straight line (row, column, or diagonal).",
|
||||
"If the board is full and there is no winner, declare a draw.",
|
||||
"Provide only the result (e.g., 'Player X wins', 'Player O wins', 'Draw').",
|
||||
],
|
||||
markdown=True,
|
||||
)
|
||||
|
||||
# Function to extract the move from the agent's response
|
||||
def extract_move(response):
|
||||
content = response.content.strip()
|
||||
match = re.search(r'\d\s*,\s*\d', content)
|
||||
if match:
|
||||
move = match.group().replace(' ', '')
|
||||
return move
|
||||
numbers = re.findall(r'\d+', content)
|
||||
if len(numbers) >= 2:
|
||||
row = int(numbers[0])
|
||||
col = int(numbers[1])
|
||||
return f"{row},{col}"
|
||||
return None
|
||||
|
||||
# Game loop
|
||||
def play_game():
|
||||
if 'current_player' not in st.session_state:
|
||||
st.session_state.current_player = player_x
|
||||
if 'symbol' not in st.session_state:
|
||||
st.session_state.symbol = "X"
|
||||
if 'move_count' not in st.session_state:
|
||||
st.session_state.move_count = 0
|
||||
|
||||
max_moves = 9
|
||||
winner = None # Initialize winner variable
|
||||
|
||||
while st.session_state.move_count < max_moves:
|
||||
# Display the board
|
||||
display_board(st.session_state.board)
|
||||
|
||||
# Prepare the board state for the agent
|
||||
board_state = get_board_state(st.session_state.board)
|
||||
move_prompt = (
|
||||
f"Current board state:\n{board_state}\n"
|
||||
f"{st.session_state.current_player.name}'s turn. Make your move in the format 'row, col'."
|
||||
)
|
||||
|
||||
# Get the current player's move
|
||||
with st.chat_message("assistant"):
|
||||
st.write(f"**{st.session_state.current_player.name}'s turn:**")
|
||||
move_response = st.session_state.current_player.run(move_prompt)
|
||||
st.code(f"Agent {st.session_state.current_player.name} response:\n{move_response.content}", language="markdown")
|
||||
move = extract_move(move_response)
|
||||
st.write(f"**Extracted move:** {move}")
|
||||
|
||||
if move is None:
|
||||
st.error("Invalid move! Please use the format 'row, col'.")
|
||||
continue
|
||||
|
||||
try:
|
||||
row, col = map(int, move.split(','))
|
||||
if st.session_state.board[row][col] is not None:
|
||||
st.error("Invalid move! Cell already occupied.")
|
||||
continue
|
||||
st.session_state.board[row][col] = st.session_state.symbol
|
||||
except (ValueError, IndexError):
|
||||
st.error("Invalid move! Please use the format 'row, col'.")
|
||||
continue
|
||||
|
||||
# Check for a winner or draw
|
||||
winner = check_winner(st.session_state.board)
|
||||
if winner:
|
||||
break # Exit the loop if there's a winner
|
||||
|
||||
# Switch players
|
||||
st.session_state.current_player = player_o if st.session_state.current_player == player_x else player_x
|
||||
st.session_state.symbol = "O" if st.session_state.symbol == "X" else "X"
|
||||
st.session_state.move_count += 1
|
||||
|
||||
# After the game loop, display the final board and involve the judge agent
|
||||
st.write("**Final Board:**")
|
||||
display_board(st.session_state.board)
|
||||
|
||||
if winner:
|
||||
st.success(f"**Result:** {winner}")
|
||||
else:
|
||||
st.success("**Result:** Draw")
|
||||
|
||||
# Judge's announcement
|
||||
st.write("\n**Game Over. Judge is determining the result...**")
|
||||
st.write("**Final board state passed to judge:**")
|
||||
st.code(get_board_state(st.session_state.board), language="markdown")
|
||||
judge_prompt = (
|
||||
f"Final board state:\n{get_board_state(st.session_state.board)}\n"
|
||||
f"Determine the winner and announce the result."
|
||||
)
|
||||
judge_response = judge.run(judge_prompt)
|
||||
st.code(f"Judge's raw response:\n{judge_response.content}", language="markdown")
|
||||
announcement = judge_response.content.strip()
|
||||
st.success(f"**Judge's Announcement:** {announcement}")
|
||||
|
||||
# Start Game Button
|
||||
if st.button("Start Game"):
|
||||
st.session_state.board = [[None, None, None],
|
||||
[None, None, None],
|
||||
[None, None, None]]
|
||||
st.session_state.current_player = player_x
|
||||
st.session_state.symbol = "X"
|
||||
st.session_state.move_count = 0
|
||||
play_game()
|
||||
else:
|
||||
st.warning("Please enter your OpenAI API key and either Deepseek/Gemini API key to start the game.")
|
||||
4
ai_agent_tutorials/ai_tic_tac_toe_agent/requirements.txt
Normal file
4
ai_agent_tutorials/ai_tic_tac_toe_agent/requirements.txt
Normal file
|
|
@ -0,0 +1,4 @@
|
|||
streamlit==1.41.1
|
||||
phidata==2.7.3
|
||||
openai==1.58.1
|
||||
google-generativeai==0.8.3
|
||||
Loading…
Reference in a new issue