app.py

import gradio as gr
from llama_index.llms import Perplexity
from llama_index.llms.base import ChatMessage

# Define the main function for handling chat interactions.
def chat_with_pplx_model(api_key, model_name, user_input, pre_prompt, system_message, temperature, max_tokens):
    # Convert max_tokens to integer if it is not empty. This controls the length of the model's responses.
    max_tokens = int(max_tokens) if max_tokens else None

    # Prepend the pre-prompt text to the user input. This allows for setting a context or instructions.
    full_user_input = f"{pre_prompt}\n{user_input}"

    # Initialize the Perplexity model with the given parameters.
    llm = Perplexity(
        api_key=api_key, 
        model_name=model_name, 
        temperature=temperature,
        max_tokens=max_tokens  # If max_tokens is None, the model uses its default value.
    )
    
    # Prepare the chat messages for interaction with the model.
    messages_dict = [
        {"role": "system", "content": system_message},  # System message, like an initial greeting or instructions.
        {"role": "user", "content": full_user_input}    # The actual user input, prepended with the pre_prompt.
    ]
    messages = [ChatMessage(**msg) for msg in messages_dict]
    
    # Get the response from the LLM.
    response = llm.chat(messages)
    return response

# Gradio Interface components.
api_key_input = gr.Textbox(label="API Key")  # Input for the API key.
model_name_dropdown = gr.Dropdown(choices=["pplx-70b-online", "pplx-7b-online", "mixtral-8x7b-instruct"], label="LLM Model Name")  # Input for the model name.
user_input = gr.Textbox(placeholder="Enter your input here", label="User Input")  # Input for user's message.
pre_prompt_input = gr.Textbox(placeholder="Enter pre-prompt here", label="Pre-Prompt")  # Input for the pre-prompt text.
system_message = gr.Textbox(placeholder="Enter system message here", label="System Message")  # Input for system message.
temperature_slider = gr.Slider(minimum=0, maximum=2, step=0.01, label="Temperature")  # Slider to adjust the temperature.
max_tokens_input = gr.Textbox(placeholder="Enter max tokens (optional)", label="Max Tokens")  # Input for max tokens.

# Creating the Gradio interface.
iface = gr.Interface(
    fn=chat_with_pplx_model,
    inputs=[api_key_input, model_name_dropdown, user_input, pre_prompt_input, system_message, temperature_slider, max_tokens_input],
    outputs="text"
)

# Launching the interface.
iface.launch(share=True)