https://github.com/flyingfathead/gpt2-tensorflow-to-pytorch-converter

Convert TensorFlow-based GPT-2 models to Pytorch in a breeze
https://github.com/flyingfathead/gpt2-tensorflow-to-pytorch-converter

conversion converter gpt-2 gpt2 llm model openai python python3 pytorch pytorch-model pytorch-models tensorflow tensorflow-model tensorflow-models tensorflow2 tf2

Last synced: about 2 hours ago
JSON representation

Convert TensorFlow-based GPT-2 models to Pytorch in a breeze

• Host: GitHub
• URL: https://github.com/flyingfathead/gpt2-tensorflow-to-pytorch-converter
• Owner: FlyingFathead
• Created: 2024-07-23T15:32:18.000Z (2 months ago)
• Default Branch: main
• Last Pushed: 2024-07-23T20:26:49.000Z (2 months ago)
• Last Synced: 2024-09-23T08:02:33.822Z (4 days ago)
• Topics: conversion, converter, gpt-2, gpt2, llm, model, openai, python, python3, pytorch, pytorch-model, pytorch-models, tensorflow, tensorflow-model, tensorflow-models, tensorflow2, tf2
• Language: Python
• Homepage: https://github.com/FlyingFathead/gpt2-tensorflow-to-pytorch-converter
• Size: 27.3 KB
• Stars: 1
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# gpt2-tensorflow-to-pytorch-converter

This repository contains a quick-to-use script to convert GPT-2 models from TensorFlow to PyTorch model format.

## Usage

1. Collect all your TensorFlow model files into a singular directory, i.e. these files:

    ```

    model-.meta

    vocab.bpe

    model-.data-00000-of-00001

    model-.index

    checkpoint

    counter

    encoder.json

    hparams.json

    ```

2. Clone the repo, install prerequisites with i.e. `pip install -r requirements.txt` if needed.

3. Run the script:

    ```bash

    python convert_model.py /path/to/your/model/files

    ```

4. The converted PyTorch model will be saved in the `./converted_model` directory.

## Notes

Have fun, I probably won't be updating this one much.

## License

This project is licensed under the MIT License.

## Contribute

All code improvements are welcome. This should at least work on all TF1.x-based GPT-2 architecture models.

## About

- Flying from the mind of [FlyingFathead](https://github.com/FlyingFathead/)

- Digital ghost code by ChaosWhisperer