Installation

Every setup installs two commands: LabGrymace (the pain tool) and LabGym_LabGrymace (our modified LabGym build).

If you already have the original LabGym installed, it is safe to install LabGym_LabGrymace. The modified build uses a different package name and command, so it will not overwrite or replace your existing LabGym installation. Both versions can coexist on the same computer.

Important

The LabGym_LabGrymace build is required — not optional. LabGrymace reads output generated only by this modified LabGym 2.9.0 build. The latest LabGym (v3.x) is not compatible. Because the two packages have different names, they can be installed side by side. See How LabGym_LabGrymace differs from the original LabGym.

Requires Python 3.10.

If Anaconda or Miniconda is installed, run conda deactivate first so the install does not go into the conda base environment. LabGrymace pins specific versions of TensorFlow and NumPy that could otherwise downgrade packages in a shared environment.

Install Python 3.10 from https://www.python.org/downloads/, then follow the steps for your system. The steps use the same method as the upstream LabGym installation, but install the modified LabGym_LabGrymace build from this repository (pip install ./LabGym_LabGrymace), not the upstream LabGym release from PyPI.

• Windows
• macOS
• Linux
• Usage

Pretrained Models

Pretrained Detector and Categorizer models are available from the LabGym Zoo.

After updating the models:

• place Detector models in LabGym_LabGrymace/LabGym_LabGrymace_copy/detectors/

• place Categorizer models in LabGym_LabGrymace/LabGym_LabGrymace_copy/models/

Then select the desired models in the LabGym_LabGrymace GUI.