Setup

Installation

The latest stable release or development release can be installed from PyPI.

Pip

Lens is on PyPi and can be installed using pip. The below will install the latest stable release of the package.

pip install credoai-lens

The latest development version can also be installed directly from github.

pip install git+https://github.com/credo-ai/credoai_lens#develop

If you are interested in contributing to the development of Lens and would like to install requirements for testing and formatting, you’ll have to install the dev requirements.

pip install credoai-lens[dev]

Conda

Lens is not on conda-forge. However, you can create a working python environment using Anaconda by following the below steps.

First, download this environment.yml file.

Then run…

conda env create --file {path-to-environment.yml}

The above will install the latest stable version of Lens. If you prefer to use the latest development version, follow the above by installing directly from github using pip (see below).

ARM Macbook installation troubleshooting

Pip installation sometimes fails on arm64 macbooks. Some packages that occasionally have issues are: pandas, scipy, scikit-learn, tensorflow, transformers. Installing these packages with the miniforge package manager seems to be the easiest way to circumvent this issue. Ensure you are using miniforge rather than anaconda! Only miniforge supports arm64 processors. E.g.:

conda install -c conda-forge -c huggingface pandas scipy scikit-learn tensorflow transformers

Warning

Tensorflow in particular is difficult to install on M1+ macbooks (macbooks using the arm64 processor). Full installation of lens (with extras) requires tensorflow. Resources below should help you get your installation working.

If your mac has an arm64 processor, it is advisable to upgrade to Monterey or later (12.3+). Doing so greatly simplifies tensorflow’s installation. In this case you can install it directly from conda running the following command:

conda install -c conda-forge tensorflow

If you are running an earlier OS (e.g., BigSur), installation can be trickier. Below are some resources that could help:

Other tutorials: