-
Notifications
You must be signed in to change notification settings - Fork 41
System requirements
Osma Suominen edited this page May 14, 2024
·
7 revisions
Annif is being developed on Ubuntu machines. However, it should work on any Linux distribution with some work. As for other OS:s, Annif itself is quite portable as it only requires a Python 3.9 (or newer) environment, although some of the optional extensions require native code libraries which may be challenging to install in some environments.
Annif is also available as a pre-built Docker image that can be run on all operating systems supporting Docker; see the Usage-with-Docker wiki page.
- At least 8GB of RAM; 16GB or more is recommended if you are planning on hosting several models, e.g. different vocabularies and/or languages.
- Annif itself requires up to 2GB disk space with all optional features installed. Corpora and models can be large, thus 100GB disk space is recommended
- Multiple CPU cores (ideally 4+) are useful, e.g. for building fastText and Omikuji models which can take advantage of parallel processing
- Home
- Getting started
- System requirements
- Optional features and dependencies
- Usage with Docker
- Architecture
- Commands
- Web user interface
- REST API
- Corpus formats
- Project configuration
- Analyzers
- Transforms
- Language detection
- Hugging Face Hub integration
- Achieving good results
- Reusing preprocessed training data
- Running as a WSGI service
- Backward compatibility between Annif releases
- Backends
- Development flow, branches and tags
- Release process
- Creating a new backend