torch>=1.5
torchvision>=0.6.0

pandas 

# testing
pytest

# Sentence Boundary Detection 
pysbd

# For visual layout detection libraries 
layoutparser>=0.2

# HF 
transformers==4.5 # Enforce the version for now 
datasets

# For extracting PDF and images from the input
pdfplumber
pdf2image

# Progress bar
tqdm

# For printing experiment results
scikit-learn