In this tutorial, we will build a chatbot using the OpenAI API and Gradio. Our chatbot will be based on the character Bart Simpson from The Simpsons and will be designed to have a mischievous and rebellious personality.
To follow along with this tutorial, you will need the following:
- A text editor like VsCode
- Python 3.6 or higher
- The OpenAI API key from openai.com
- Git (optional)
Setting up the Project
First, let's create a new project directory and set up our virtual environment:
mkdir chatbot cd chatbot python -m venv env source env/bin/activate
Next, let's install the required dependencies:
pip install openai gradio
Writing the Code
Now, let's write the code for our chatbot. We will start by creating a file named gpt.py and adding the following code:
- The prompt
- get_response function that leverages the OpenAi Api to query the Davinci model and returns the answer in text.
import openai import gradio as gr PROMPT = """The following is a conversation with Bart Simpson. As Bart Simpson, I would describe myself as a mischievous, rebellious, and adventurous kid. I'm always getting into trouble and finding new ways to have fun and cause chaos. Me: Hello, who are you? Bart: Hey, it's great to hear from you! Not much has changed here in Springfield. What about you, what have you been up to? Me: """ def get_response(prompt): """ Function that generates a response from the OpenAI API based on a given prompt. Parameters: - prompt (str): The prompt to generate a response for. Returns: - response (str): The response from the OpenAI API. """ response = openai.Completion.create( model="text-davinci-003", prompt=PROMPT, temperature=0.9, max_tokens=150, top_p=1, frequency_penalty=0, presence_penalty=0.6, stop=[" Me:", " Bart:"] ) return response.choices.text
In this part, create a file named main.py and add the following code:
- Cloning_gpt function that generates a response based on a given input and conversation history.
- Opening the Gradio block that will give us the interface to interact with the chat bot.
import gradio as gr from gpt import get_response def cloning_gpt(input, history): ''' Function that generates a response based on a given input and conversation history. Parameters: - input (str): The input to generate a response for. - history (list): A list of tuples containing previous inputs and outputs in the conversation. Returns: - history (list): The updated conversation history with the new input and output appended. - history (list): A copy of the updated conversation history. ''' history = history or  # if history is None, set it to an empty list past = list(sum(history, ())) # flatten the history list past.append(input) # add the current input to the history inputs = ' '.join(past) # join the history into a string output = get_response(inputs) # get the model's response history.append( (input, output)) # add the current input and output to the history return history, history # return the history as both the output and the state block = gr.Blocks() with block: gr.Markdown("""<h1><center>ChatGpt "Bart version" </center></h1>""") chatbot = gr.Chatbot() message = gr.Textbox(placeholder="Hey I'm Bart Simpson, ask me anything!") state = gr.State() submit = gr.Button("SEND") submit.click(cloning_gpt, inputs=[message, state], outputs=[chatbot, state]) block.launch( debug=True, share=True ) # debug=True to run locally and share=True to share the app publicaly
Laucnh the project
When launching the app, you can choose
debug=True to run locally or
share=True to share the app publicaly or both. USually the it can be accessible publicly for 72 hours.
Check out the Github repo
Top comments (0)