Large-scale Self-supervised Pre-training Across Tasks, Languages, and Modalities

This repository contains demos I made with the Transformers library by HuggingFace.

A Repo For Document AI

The MERIT Dataset is a fully synthetic, labeled dataset created for training and benchmarking LLMs on Visually Rich Document Understanding tasks. It is also designed to help detect biases and improve interpretability in LLMs, where we are actively working. This repository is actively maintained, and new features are continuously being added.

All in one package for Document (image, pdf) Classification. Unified Interface for google ocr and tesseract. Train, evaluate, and infer using fasttext, Small language models (NER), Small Vision Language Models (layoutlm), and LLM.

LayoutLMv3 applied to a VQA problem with infographics.

Create meta data from resume

Exploring LayoutLM for Smart OCR Capabilities

Fine-tuning LayoutLMv3 on the SROIE dataset to build a receipt understanding model

Document Image Understanding: Analysis of 2 datasets

AdminDoc-X is an AI-powered document intelligence platform for extracting, classifying, and structuring information from administrative documents using OCR, LayoutLM, and NLP.

SamvidAI — Enterprise Contract Intelligence powered by OpticalRAG Multimodal document understanding system for clause extraction, legal risk scoring, and explainable contract analysis using layout-aware RAG pipelines.

Mini Projects that are developed using Python.

LayoutLMv3 for invoice, contract, and form parsing with key-value extraction

Prototypical Networks for Information Extraction in Visual Documents. Weights can be found at https://drive.google.com/file/d/1Zrp7QaZIf0H_FFRx_LhB0uZTqDUSis2H/view?usp=sharing.

Automated invoice data extraction using LayoutLMv3 (PyTorch) with PyQt6 annotation tool

A fully automated document scanning, OCR processing, storage, and search pipeline built around the ScanSnap iX1300 scanner.