NLP Text summarization task

Created a specialized annotated dataset for supervised learning and used the annotated dataset for the fine-tuning the BERT model, focusing on a regression task by performing extractive text summarization of financial data.

Run Locally (It is preferred to run on Google colab with GPU.)

Clone the project (or download manually)

  git clone https://github.com/Mahendranath94/BERT_Pipeline

Go to the project directory, place all the annual reports in SourceData folder company wise (refer to the files structure). Make sure the folders containing company wise annual reports are named as "Annual Reports" or "Annual_Reports".

Run BERT-TS.ipynb file

Follow the below files structure:

Make sure there are no empty .txt file in Annual Reports folders.