VieTable donut docvqa demo

ProviderYuukiAsuna
Categorydocument-question-answering
Licensemit
Downloads34
Stars0

Overview

The VieTable Donut DocVQA model is a specialized document visual question answering tool designed to extract structured information from unstructured images without relying on traditional OCR engines. By leveraging an end-to-end transformer architecture, it maps visual inputs directly to text sequences, making it particularly effective for parsing tables, forms, and complex document layouts where spatial positioning is critical. For developers, this means a simplified pipeline—reducing the latency and error propagation typically associated with separate OCR and NLP stages. It is an ideal choice for automating data entry, invoice processing, and digital archiving tasks where precision in document-based retrieval is paramount.

Highlights

  • OCR-free end-to-end visual question answering
  • High precision in table and form parsing
  • Reduced pipeline latency by eliminating separate OCR
  • MIT licensed for flexible commercial integration

Usage

Install
# Install Hugging Face transformers
pip install transformers torch
SDK Usage
# Load model with transformers
from transformers import AutoModel, AutoTokenizer

model = AutoModel.from_pretrained("YuukiAsuna/VieTable-donut-docvqa-demo")
tokenizer = AutoTokenizer.from_pretrained("YuukiAsuna/VieTable-donut-docvqa-demo")

Hugging Face Download

We recommend downloading the model via the Hugging Face CLI or Hub SDK.

Guidance：Before downloading, install huggingface_hub with:

Guidance 
pip install -U huggingface_hub

CLI Download

Download the full repository

Download the full repository 
huggingface-cli download YuukiAsuna/VieTable-donut-docvqa-demo

Download a single file to a local folder (e.g. config.json into ./dir)

Download a single file to a local folder (e.g. config.json into ./dir) 
huggingface-cli download YuukiAsuna/VieTable-donut-docvqa-demo config.json --local-dir ./dir

See the official docs for more CLI options

SDK Download

SDK Download 
# 模型下载
from huggingface_hub import snapshot_download
model_dir = snapshot_download('YuukiAsuna/VieTable-donut-docvqa-demo')

Git Download

Make sure git-lfs is installed first

Git Download 
git lfs install
git clone https://huggingface.co/YuukiAsuna/VieTable-donut-docvqa-demo

To skip LFS large-file downloads, use:

Skip LFS 
GIT_LFS_SKIP_SMUDGE=1 git clone https://huggingface.co/YuukiAsuna/VieTable-donut-docvqa-demo

Model files are hosted on the Hugging Face Hub — download directly via HF CLI / SDK / Git, not through this site.

PyTorch / Transformers Usage

Install Transformers

Install Transformers 
pip install -U transformers torch

Load the model and run inference

Load the model and run inference 
from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained('YuukiAsuna/VieTable-donut-docvqa-demo')
tokenizer = AutoTokenizer.from_pretrained('YuukiAsuna/VieTable-donut-docvqa-demo')

Full Documentation

来源: HuggingFace

---
library_name: transformers
license: mit
datasets:

  • YuukiAsuna/VietnameseTableVQA

language:
  • vi

base_model:
  • naver-clova-ix/donut-base

pipeline_tag: document-question-answering
---

Model Card for Model ID

<!-- Provide a quick summary of what the model is/does. -->

VieTable Donut DocVQA is a fine-tuned version of the Donut model for the Vietnamese DocVQA (Table data)

BibTeX entry and citation info

bibtex
@article{DBLP:journals/corr/abs-2111-15664,
  author    = {Geewook Kim and
               Teakgyu Hong and
               Moonbin Yim and
               Jinyoung Park and
               Jinyeong Yim and
               Wonseok Hwang and
               Sangdoo Yun and
               Dongyoon Han and
               Seunghyun Park},
  title     = {Donut: Document Understanding Transformer without {OCR}},
  journal   = {CoRR},
  volume    = {abs/2111.15664},
  year      = {2021},
  url       = {https://arxiv.org/abs/2111.15664},
  eprinttype = {arXiv},
  eprint    = {2111.15664},
  timestamp = {Thu, 02 Dec 2021 10:50:44 +0100},
  biburl    = {https://dblp.org/rec/journals/corr/abs-2111-15664.bib},
  bibsource = {dblp computer science bibliography, https://dblp.org}
}
View official source
Join our Telegram