To install Label Sleuth, your system should satisfy the following requirements:
Operating System: Windows, MacOS, or Linux
Hardware: At least 2 CPU cores and 4GB RAM (for improved performance, 8 CPU cores and 16GB RAM are recommended, especially when working with large datasets). Note that some models may have additional hardware requirements, as outlined in the models documentation.
Recommended procedure to install Label Sleuth:
Anaconda will allow you to create a separate Python environment for Label Sleuth. This is important as Label Sleuth currently only supports Python 3.8 (other versions may cause issues). Download and install Anaconda from its official web-site:
Open a new terminal or restart it if it is already open.
In the new terminal, create a Python environment for Label Sleuth:
conda create --yes -n label-sleuth python=3.9
Then, activate the new environment:
conda activate label-sleuth
Install Label Sleuth
Install Label Sleuth by running the following command:
pip install label-sleuth
Fire it all up
Start Label Sleuth:
python -m label_sleuth.start_label_sleuth --load_sample_corpus wiki_animals_2000_pages
This command also pre-loads a collection of Wikipedia documents that will be used in the tutorial below.
Access Label Sleuth on your browser by navigating to the following page:
Follow tutorial (Recommended)
Now that you have installed Label Sleuth, we strongly recommend following our step-by-step tutorial to get acquainted with the system.
Changing the project directory and port
When starting up Label Sleuth, the project directory and port can be customized by appending to the startup command
python -m label_sleuth.start_label_sleuth the following flags:
Set location of project files to
Set port to
<port_number>(default is 8000)
Set host to
<ip_or_name>(default is localhost, use 0.0.0.0 to expose the service to external communication)