deepv_api_tutorial.py

from gradio_client import Client
import os

# This will be the URL of DeepV's Gradio App
GRADIO_APP_URL = "FICS-LLM/DeepV"  # Replace with your actual app URL

try:
    # Initialize the Gradio client
    client = Client(GRADIO_APP_URL)
    print("DeepV Gradio client initialized successfully!")

    # input design spec to prompt LLM
    design_spec = "Create me a 2x1 multiplexer with the following module header: module mux2x1(input wire a, input wire b, input wire sel, output wire y);"

    # number of retrieved documents for RAG
    k_value = 1

    # Set if you would like to use DeepV's RAG technique
    rag_enabled = True

    # Select your model: gpt-4o, gpt-4.1, gpt-5-chat-latest
    model_name = "gpt-4o"

    openai_key = "" # Replace with your actual OpenAI API key
    # Note: Your OpenAI API key is only active for this session and is not saved or stored. 

    # Customize some generation parameters (temperature, top_p, max_tokens)
    temp_value = 0.5
    top_p_value = 0.9
    max_tokens = 1024

    # --- Call the Gradio app's API endpoint ---
    # The `predict` method corresponds to a component's action.
    response = client.predict(
        design_spec,
        rag_enabled,
        k_value,
        model_name,
        openai_key,
        temp_value,
        top_p_value,
        max_tokens,
        api_name="/run_generation" # Assuming 'run_generation' is exposed as an API
    )

    # --- Process the response ---
    
    generated_code = response[0]

    print("\n--- API Call Successful! ---")
    print("\nGenerated Verilog Code:")
    print("-------------------------")
    print(generated_code)

except Exception as e:
    print(f"An error occurred: {e}")
    print("Please ensure your Gradio app is running and the URL is correct.")
    print("You might also need to find the correct `fn_index` if the API call fails.")