GPT2 XY

This repository uses HuggingFace's GPT2 Implementation and exposes an creates a nice user interface for testing GPT2 power. Even though it may not be exactly as good as authors' original tensorflow implementation, it still surprises sometimes.

Update: Only GPT-2 Small (117M paramter) model is supported on Cloud Run. Medium/Large or XL cannot be run on Cloud Run due to memory limits.

Get Started

To get started on this project, follow these steps:

• Clone: git clone https://github.com/NaxAlpha/gpt-2xy.git and cd gpt-2xy
• Install Dependencies: pip install -r requirements.txt
• Download the gpt related data: python download.py
• Run server: python main.py
• Open Web Browser and visit: http://localhost:8080/
• Type Microsoft and Google and press GENERATE TEXT
• Enjoy!

You may also want to build and run this project in docker container

• Build: docker build -t gpt-2xy . (will take some time)
• Run: docker run -p 8080:8080 --rm -d gpt-2xy

If successful, you will see something like this:

Full Tutorial

This repository contains code for my medium tutorial on Deploy Machine Learning Model in Google Cloud using Cloud Run