Installation¶

Hyperbase requires Python >=3.10.

Install with pip¶

pip install hyperbase

Install with uv¶

uv is a fast Python package manager. To add hyperbase to your project:

uv add hyperbase

Or to install it in a standalone environment:

uv pip install hyperbase

Install from source¶

Clone the repository and install:

git clone https://github.com/hyperbase/hyperbase.git
cd hyperbase
pip install .

Or using uv:

git clone https://github.com/hyperbase/hyperbase.git
cd hyperbase
uv sync

To run the tests from the project root:

pytest

Or using uv:

uv run pytest

Parsers¶

Hyperbase uses a plugin architecture for parsers. The core package does not include any parser, it only specifies the abstract parser interface. Parsers are installed separately as Python packages that register themselves via entry points.

Package	Plugin name	Description	License
`hyperbase-parser-ab`	`alphabeta`	AlphaBeta parser using spaCy	Open source (MIT)
`hyperbase-parser-gen`	`generative`	Multilingual generative parser based on fine-tuned transformer model	Proprietary

AlphaBeta is the "classical" parser developed for Semantic Hypergraphs. It is based on spaCy and depends on its language models, that have to be installed to support a given language. It is released under the same permissive open source license as hyperbase (MIT).

Generative is the "modern" multilingual parser, which produces high-quality parses for many languages. It is based on the fine-tuning of a more powerful transformer model and requires a GPU to run with acceptable speed. We plan on making this parser available for researchers and under a commercial license otherwise. Contact us if you are a researcher and wish to have early access to this parser.

Installing a parser¶

Install a parser with pip:

pip install hyperbase-parser-ab

Or with uv:

uv add hyperbase-parser-ab

AlphaBeta language support: installing spaCy models¶

For example, for English language support:

python -m spacy download en_core_web_trf

Or with uv:

uv run python -m spacy download en_core_web_trf

Replace with the appropriate model from the table below.

Language	Code	Model
Chinese	`zh`	`zh_core_news_trf`
English	`en`	`en_core_web_trf`
French	`fr`	`fr_dep_news_trf`
German	`de`	`de_dep_news_trf`
Italy	`it`	`it_core_news_lg`
Portuguese	`pt`	`pt_core_news_lg`
Spanish	`es`	`es_dep_news_trf`

The table shows the languages that are currently configured. In principle, any language for which a spaCy model is available can be supported. If you are interested in adding a language and can verify accuracy, please add a pull request or open an issue in the plugin's repository.

Verifying installation¶

You can list all installed parsers with the CLI:

hyperbase parsers

uv run hyperbase parsers

Using parsers¶

Once installed, parsers can be used from the interactive REPL:

hyperbase repl --parser alphabeta --lang en

uv run hyperbase repl --parser alphabeta --lang en

Or programmatically:

from hyperbase.parsers import get_parser

parser = get_parser("alphabeta", lang="en")
result = parser.parse_text("The sky is blue.")