The objective of this project fine-tune the pre-trained Transformer Decoder-based language GPT2 models to obtain a very powerful abstractive text summarizer.
install from the requirements.txt
pip install -r requirements.txt
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'
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
This folder contains colab notebook that guide you through the summarization by GPT-2. You should be able to 