Running An Experiment From A Different Repository

If you have your own repository and want to run your own experiment or a benchmark, you do not need to replicate the tbp.monty benchmarks setup.

📘
We have a tbp.monty_project_template template repository, so that you can quickly use tbp.monty for your project, prototype, or paper.

First, you will need to install tbp.monty as a dependency using conda. You will need to add thousandbrainsproject to your list of channels, as well as the specific tbp.monty dependency:

# environment.yaml

# ...
channels:
  # ...
  - thousandbrainsproject

dependencies:
  # ...
  - thousandbrainsproject::tbp.monty
  - pip
  - pip:
    - -e .[dev]

You'll also want to add tbp.monty and a Hydra dependency to your pyproject.toml:

[project]
# ...
dependencies = [
    # ...
    "tbp.monty", # imported via conda (thousandbrainsproject::tbp.monty)
    "hydra-core>=1.3.2", # Hydra is used for configuring Monty
]

With the enviroment.yaml configured, create your environment:

conda env create -f environment.yaml

📘
tbp.monty is only distributed for x86_64 architecture, so if you are on Apple Silicon, your conda environment needs to be created using --subdir=osx64 option:
conda env create -f environment.yaml --subdir=osx-64

Once your conda environment is configured, the general setup is:

from tbp.monty.frameworks.run_env import setup_env

# call setup_env() to initialize environment used by
# tbp.monty configuration and runtime.
setup_env()

# import run to run the experiment
from tbp.monty.frameworks.run import main  # noqa: E402

if __name__ == "__main__":
    main()

from tbp.monty.frameworks.run_env import setup_env

# call setup_env() to initialize environment used by
# tbp.monty configuration and runtime.
setup_env()

# import run_parallel to run the experiment, parallelizing across episodes
from tbp.monty.frameworks.run_parallel import main  # noqa: E402

if __name__ == "__main__":
    main()

The above scripts let you run the existing tbp.monty experiments from your own project. For example:

python run.py experiment=tutorial/first_experiment

To run your own experiment, you need to configure it using Hydra (hydra-core>=1.3.2). For the purpose of this tutorial, we'll assume that your Hydra configuration is stored in src/project/conf:

src
`- project
   `- conf
      `- experiment
         `- example.yaml
environment.yaml
pyproject.toml
run.py
run_parallel.py

With the above setup in place and a correct configuration present in example.yaml, you can now execute it via:

python run.py experiment=example

python run_parallel.py experiment=example

For a full working example, see the template repository: https://github.com/thousandbrainsproject/tbp.monty_project_template.

Help Us Make This Page Better

All our docs are open-source. If something is wrong or unclear, submit a PR to fix it!

Make a Contribution

Learn how to contribute to our docs