README.md

1. Fine-tuning on Document AI Tasks

This repo has my implementation of models used for fine-tuning on document AI task.

1. Finetuning LayoutLMV3 on FUNSD Dataset
2. Finetuning LiLT on FUNSD Dataset
3. Finetuning RoBERTa on FUNSD Dataset
4. Finetuning DocFormer on FUNSD Dataset

Current Working:

• Currently working on finetuning models on FUNSD Dataset

WANDB Report:

FUNSD Dataset:

• The wandb progress report can be found out here here

2. Paper Implementations:

Natural Language Processing

• DocFormer: End-to-End Transformer for Document Understanding [Paper] [Code]
• LiLT: A Simple yet Effective Language-Independent Layout Transformer,for Structured Document Understanding [Paper] [Code]
• EIT: Enhanced Interactive Transformer [Paper][Code]

Visual Question Answering

• LaTr: Layout-aware transformer for scene-text VQA [Paper] [Code]

Computer Vision

• Vision Image Transformer (ViT) [Paper] [Code]

3. Projects:

• Medical Visual Question Answering
• Customer Segmentation (Unsupervised Learning)
• Brain MRI Segmentation (Classification and Segmentation)
• Image Classification (Cancer detection)
• Movie Recomendation using Cosine Similarity
• Speech Recognition