rohitashwa1907/text-summarization-using-gpt2

Train a GPT2 model to summarize long text (<600 words).

Description

The objective of this project fine-tune the pre-trained Transformer Decoder-based language GPT2 models to obtain a very powerful abstractive text summarizer.

setting up the environment

install from the requirements.txt

pip install -r requirements.txt

Training the GPT2

mkdir fine_tuned_folder

python train_command_line.py --epochs=1 --data_path='insert-path-to-training-data-here' --model_arch_name='name-of-the-gpt2-model' --model_directory='fine_tuned_folder'

Generating Summaries

python eval.py --input_file='insert-path-to-text-data-here' --model_directory='insert-path-to-pretained-model-here' --model_arch_name='name-of-the-gpt2-model' --num_of_samples='num-of-samples-to-generate

Notebook

This folder contains colab notebook that guide you through the summarization by GPT-2. You should be able to , and play with your data.