app.py

import streamlit as st
import json
from transformers import pipeline

@st.cache_resource
def load_model(model_name):
    return pipeline("text-generation", model=model_name)

def main():
    st.title("Prebid Module Generator")
    st.write("Enter a Prebid module, such as 'appnexusBidAdapter', and get a generated Prebid installed module output starting from that setting onward. Using '[' will generate common Prebid modules from the beginning. The model currently has a capped output of 1000 characters.")

    st.subheader("Intended Uses")
    st.write("This model is designed to assist publishers in understanding and exploring what modules most publishers use with their Prebid set-up. It can serve as a valuable reference to gain insights into common Prebid modules, best practices, and different approaches used by publishers across various domains. The model should be seen as a helpful tool to gain inspiration and understanding of common Prebid modules but not as a substitute for thorough testing and manual review of the final modules used.")
    st.write("To learn more about the default model, visit the [Prebid_Module_GPT2 model page](https://huggingface.co/PeterBrendan/Prebid_Module_GPT2). You can also refer to the [official Prebid Documentation on pbjs.setConfig](https://docs.prebid.org/dev-docs/publisher-api-reference/setConfig.html) for more information.")

    st.write("*Note:* The model may take some time to generate the output.")

    # Create a text input field for user prompt
    user_input = st.text_input("Enter a Prebid module:", "")

    # Check if the user input is empty
    if user_input:
        # Select the model
        model_name = "PeterBrendan/Prebid_Module_GPT2"

        # Load the Hugging Face model
        generator = load_model(model_name)

        # Display 'Generating Output' message
        output_placeholder = st.empty()
        with output_placeholder:
            st.write("Generating Output...")

        # Generate text based on user input
        generated_text = generator(user_input, max_length=1000, num_return_sequences=1)[0]["generated_text"]

        # Clear 'Generating Output' message and display the generated text
        output_placeholder.empty()
        st.write("Generated Text:")
        try:
            parsed_json = json.loads(generated_text)
            beautified_json = json.dumps(parsed_json, indent=4)
            st.code(beautified_json, language="json")
        except json.JSONDecodeError:
            st.write(generated_text)

# Run the app
if __name__ == "__main__":
    main()