Skip to content

Getting started

Already using GROBID? Read the upgrade guide first

If you are upgrading an existing installation from an earlier version, start with the Upgrade guide — GROBID 0.9.0 introduces build-environment, runtime and model changes that require attention before pulling the new version.

TL;DR

  • Docker: The easiest way to run Grobid is via Docker on Linux, Mac and Windows.
  • Clients: Grobid is written in Java, but, fear not, various clients in Python and other languages are available. See here for more information.
  • Consolidation: Grobid can merge and reconcile extracted bibliographic data and citations using Crossref. With great power comes great responsibility, read here before using consolidation. Consolidation has impact on performances.
  • Deep Learning & GPU: Grobid’s best accuracy relies on deep learning models, which benefit from GPU acceleration. For fulltext-only extraction, use the CPU-only lightweight image.
  • Production configuration: See here for tuning Grobid for production use.

Tip

Keep reading until the end of the page.

Using Grobid from the cloud

The simpler way to play with Grobid is to use the Grobid instance or Mirror Grobid instance where a light instance of Grobid is deployed. You can use it to process a PDF file, or to test the Grobid web service. The space is free and does not require any authentication.

Grobid space is for demonstration only

This grobid space is not intended for production use, it is only a demonstration of Grobid capabilities. For production use, please deploy a local version or contact us.

Quick AWS grobid deployment

For quickly spin up and down a Grobid instance on Amazon Web Service EC2, you can check out the aws-grobid project.

Running Grobid locally

Tip

The standard way to run Grobid locally is to use Docker for starting a Grobid server.

For installing Docker on your system, please visit the official Docker documentation here.

Grobid docker images are available on both at Docker Grobid Hub (grobid/grobid) and Docker lfoppiano Hub (mirror) (lfoppiano/grobid) repositories using the same tag naming conventions.

Changes of the docker image naming conventions

The lfoppiano/grobid without any suffix (e.g. lfoppiano/grobid:0.7.2) was used for the CRF image. Starting from version 0.8.2 (included), the tag without any suffix will refer to the full image. To avoid problems, please use the grobid/grobid repository instead.

For convenience, we provide two Grobid docker images:

  • the full image (docker tag {version}-full, e.g. grobid/grobid:0.9.0-full) provides the best accuracy, because it includes all the required Python and TensorFlow libraries, GPU support and all Deep Learning model resources. However, it requires more resources, ideally a GPU (it will be automatically detected on Linux). If you have a limited amount of PDF, a good machine, and prioritize accuracy, use this Grobid flavor. To run this version of Grobid, the command is: 
docker run --rm --gpus all --init --ulimit core=0 -p 8070:8070 grobid/grobid:0.9.0-full
  • the lightweight image (docker tag {version}-crf, e.g. grobid/grobid:0.9.0-crf) offers the best runtime performance (runs on CPUs only), memory usage and Docker image size. However, it does not use some of the best performing models in terms of accuracy. If you have a lot of PDF to process, a low resource system, and accuracy is not so important, use this flavor:
docker run --rm --init --ulimit core=0 -p 8070:8070 grobid/grobid:0.9.0-crf

More documentation on the Docker images can be found here.

From there, you can check on your browser if the service works fine by accessing the welcome page of the service console, available at the URL http://localhost:8070. The GROBID server can be used via the web service.