BFTM-MODEL

The BFTM (Budget-Focused Text Model) is an advanced NLP model specifically designed for extracting, processing, and analyzing budget-related information from unstructured text data. Leveraging cutting-edge unsupervised cross-lingual representation learning, as introduced by Conneau et al.[1], the model demonstrates exceptional performance in identifying and categorizing financial allocations, expenditures, and related data points across multiple languages.

Model Access

The BFTM model is available for download and integration via Google Drive:

BFTM-MODEL -> AI-Application -> Models

Ensure that the model is placed in the same directory as the Notebooks and Datasets for optimal functionality. Download the whole Models folder.

Installation

This project requires the setup of two separate environments to accommodate the different dependencies for notebook execution and model operations.

PDF Scraper and Text Extraction

For notebooks related to PDF scraping and text extraction, please install dependencies from pdf_related_requirements.txt:

pip install -r pdf_related_requirements.txt

Model Fine-tuning, Evaluation, and Application

For model fine-tuning, evaluation, and running the application, dependencies should be installed from model_requirements.txt:

pip install -r model_requirements.txt

Usage

After setting up the required environments and dependencies, the notebooks can be executed to perform PDF scraping, text extraction, and further processing with the BFTM model.