Text2LaTeX

A deep learning project that converts mathematical equations from images to LaTeX code using a combination of image encoding and transformer-based variational autoencoder (VAE) architecture.

Project Structure

• image_to_latex.py: Main inference script for converting images to LaTeX
• train_image_encoder.py & train_image_encoder_rms.py: Training scripts for the image encoder component
• train_latex_vae.py: Training script for the LaTeX VAE model
• AutoEncoder_Latex.py: Core model architecture definitions
• latex_to_latex.py: Utility for LaTeX processing
• inspect_loss.py: Tool for analyzing training losses
• models: Directory with model weights for encoder-decoder = project_report: contains pdf containing the project report

Contributers

• Parin Arora
• Vishune Varadrajaran