Donut Model

The Donut model was proposed in OCR-free Document Understanding Transformer https://huggingface.co/docs/transformers/model_doc/donut

Overview

Donut uses Variational Autoencoders (VAEs) and Monte Carlo sampling to detect anomalies in time-series data. It's particularly effective for KPIs (Key Performance Indicators) in business systems.

Usage

Usage tips The quickest way to get started with Donut is by checking the tutorial notebook, which show how to use the model at inference time as well as fine-tuning on custom data. Donut is always used within the VisionEncoderDecoder framework.

Training

We refer to the tutorial notebooks.