https://github.com/cyberzhg/keras-gpt-2

Load GPT-2 checkpoint and generate texts
https://github.com/cyberzhg/keras-gpt-2

gpt-2 keras language-model nlp

Last synced: about 1 month ago
JSON representation

Load GPT-2 checkpoint and generate texts

• Host: GitHub
• URL: https://github.com/cyberzhg/keras-gpt-2
• Owner: CyberZHG
• License: mit
• Archived: true
• Created: 2019-02-18T06:33:45.000Z (over 5 years ago)
• Default Branch: master
• Last Pushed: 2022-01-22T11:44:25.000Z (almost 3 years ago)
• Last Synced: 2024-09-27T13:23:14.400Z (about 1 month ago)
• Topics: gpt-2, keras, language-model, nlp
• Language: Python
• Homepage: https://pypi.org/project/keras-gpt-2/
• Size: 94.7 KB
• Stars: 128
• Watchers: 7
• Forks: 32
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Keras GPT-2

[![Version](https://img.shields.io/pypi/v/keras-gpt-2.svg)](https://pypi.org/project/keras-gpt-2/)

![License](https://img.shields.io/pypi/l/keras-gpt-2.svg)

\[[中文](https://github.com/CyberZHG/keras-gpt-2/blob/master/README.zh-CN.md)|[English](https://github.com/CyberZHG/keras-gpt-2/blob/master/README.md)\]

Load pretrained weights and predict with [GPT-2](https://d4mucfpksywv.cloudfront.net/better-language-models/language_models_are_unsupervised_multitask_learners.pdf).

## Install

```bash

pip install keras-gpt-2

```

## Demo

```python

import os

from keras_gpt_2 import load_trained_model_from_checkpoint, get_bpe_from_files, generate

model_folder = 'xxx/yyy/117M'

config_path = os.path.join(model_folder, 'hparams.json')

checkpoint_path = os.path.join(model_folder, 'model.ckpt')

encoder_path = os.path.join(model_folder, 'encoder.json')

vocab_path = os.path.join(model_folder, 'vocab.bpe')

print('Load model from checkpoint...')

model = load_trained_model_from_checkpoint(config_path, checkpoint_path)

print('Load BPE from files...')

bpe = get_bpe_from_files(encoder_path, vocab_path)

print('Generate text...')

output = generate(model, bpe, ['From the day forth, my arm'], length=20, top_k=1)

# If you are using the 117M model and top_k equals to 1, then the result will be:

# "From the day forth, my arm was broken, and I was in a state of pain. I was in a state of pain,"

print(output[0])

```