Quadrillion IO

Follow

--

Bard is a large language model chatbot developed by Google AI. It is gaining more traction due to its speed and internet capabilities in comparison to Open AI’s ChatGPT. Google is yet to release an official API for developers to integrate Bard into their applications. However, there have been numerous projects that have built unofficial Python based APIs for Bard.

This article will guide you through building a replicate of Bard by using an unofficial Bard API and Streamlit. By following the steps in this article, you will be able to create a chatbot that can answer questions and generate text.

Note: This article assumes you are familiar with building Python applications in VS Code. If you need help with getting started, check out this guideline:

code.visualstudio.com

Create a new project with a virtual environment in VS Code. Then install the following libraries:

To use the unofficial Bard API, we will first have to get a secure token that for the purposes of this article will be called the Bard API Key. Open Chrome and press F12 to trigger the development console. Next, visit https://bard.google.com/ and follow the steps below to get the Bard API Key:

Now that we have the key, let’s store it in a configuration file. In your VS Code Project, create a new file called config.json then add the following code:

Note: You can skip this step and directly embed your API key into the code, but that goes against best practices.

developers.google.com

We are ready to code our own chatbot! Create a new file called app.py and import these libraries.

First, we will have to read the Bard API Key from our configuration file and store it in a variable. Then we will initialize the Bard API by passing the key’s variable.

bard_api_key = data["BARD_API_KEY"]

bard = Bard(token=bard_api_key, timeout= 30)

Now that we have initialized the API, let’s create a function that takes a prompt as an input and calls the Bard API to get a response.

Notice how we include the history of our conversation into the prompt, so Bard has the full context of our discussion? Let’s create a function to clear the history so we can start a new discussion.

Lastly, let’s create another function that handles user input and passes it to Bard for responses.

We are almost done! The last step is to create the user interface for our web application using the code below:

# create a button to clear the session state (chat history)st.button("New Topic", on_click=on_btn_click)if 'bard_responses' not in st.session_state:st.session_state['bard_responses'] = []if 'user_responses' not in st.session_state:st.session_state['user_responses'] = []

# create the text input fieldwith st.container():st.text_input("User Response:", on_change=on_input_change, key="user_input")

# show a loading spinner while the bard api is getting the response# then show the chat historywith st.spinner("Loading..."):if st.session_state['bard_responses']:for i in range(len(st.session_state['bard_responses'])-1, -1, -1):message(st.session_state["bard_responses"][i], key=str(i))message(st.session_state['user_responses'][i], is_user=True, key=str(i) + '_user')

Note: You can grab the full code on GitHub:

github.com

In order to run our web application in VS Code, we first need to create the appropriate launch configuration:

4. Enter “streamlit” for the module name

5. Add the following line of code to the launch.json file: , “args”: [“run”, “${file}”],

Note: Your launch.json file should looke like this.

6. Save the file

Running Our Chatbot

Now, we are ready to run our code! Let’s give it a try.

Let’s also confirm that Bard is able to recall our conversation history.

Great! Now, it’s your turn to experiment with your very own Bard Chatbot.

Here are some questions to try: