Week 9 Mini-Project

Developed by

Zach Xing

Project Specifications

Construct a web interface using Streamlit.
Integrate with a freely accessible LLM (from Hugging Face).
Make the model accessible online through either Streamlit or another comparable service.

Application Description

Capabilities

This application, designed using Streamlit, facilitates English translations from Chinese text. It leverages an open-source LLM, specifically the Helsinki-NLP/opus-mt-zh-en model from the transformers library. Users can test the functionality by inputting a sentence, pressing the "translate" button, and viewing the English translation displayed below the input area.

Access Link

The application is hosted on Streamlit at the following URL: https://zs148-ids721-week9-btkkfloe3a4qjpqtfgf6mn.streamlit.app/

Implementation Steps

Required Installations

Before launching the app, install the following essential packages:

sudo pip install streamlit transformers tensorflow sentencepiece

Building the Streamlit Application

Develop a Python script to perform the translation task:

translator = pipeline("translation", model="Helsinki-NLP/opus-mt-zh-en")

text_to_translate = st.text_area("Please type input: (Example: 你好，很高兴认识你)", value='', height=250, max_chars=500)

if st.button('Translate'):
    if text_to_translate:
        translation = translator(text_to_translate, max_length=400)[0]['translation_text']
        st.write("Translated result:", translation)
    else:
        st.write("Please enter some text to translate.")

To run a local test of the app, execute:

sudo streamlit run streamlit_app.py

For adapting the app to different translation tasks, consult the Hugging Face model repository.

Deployment Through Streamlit

To prepare for deployment, compile all necessary dependencies into a requirements.txt file. After uploading the local project to a GitHub repository, link the repository through your Streamlit account for deployment.

zihanxing / simple-streamlit-app