np_GPT2

np_GPT2 is a non-sampling implementation of GPT-2 using only Numpy for tensor operations.

Overview

This project aims to replicate the functionality of GPT-2 using Numpy, without relying on sampling mechanisms. It provides a lightweight alternative for understanding and working with GPT-2.

Running the Comparison Script

To compare the outputs of different GPT-2 implementations, run the got2_comparison.py script with Python 3.12. The script will display responses from:

• GPT-2 directly from the HuggingFace transformer library
• GPT-2 from a debugging fork of the HuggingFace transformer library
• np_GPT2 using only Numpy

The goal is for all responses to match, demonstrating the correctness of the np_GPT2 implementation.

Requirements

• Python 3.12
• Numpy
• HuggingFace Transformers library (for comparison purposes)
• HuggingFace Transformers fork (for debug purposes)

Installation

1. Clone the repository:

   git clone https://github.com/RyanTomich/np_GPT2.git