Install_locally.md

Install Ray and required libraries locally

Here, you can find instructions about how to locally install Ray and all libraries required for Ray Saturday.

Follow the instructions to set it up.

Clone this repository

In your terminal run:

git clone https://github.com/ray-project/ray-saturday-dec-2022.git

Go to the repository directory. In your terminal run:

cd ray-saturday-dec-2022

Instructions for Unix users

You should install dependencies in a new virtual environment. If you are not familiar with virtualenv, consult their documentation before you move forward.

Make sure that you have virtual env with python==3.9.12.

Once in the virtualenv, run in your terminal:

pip install -r environment/requirements.txt

Make sure to install PyTorch==1.13.0 that is suitable for your system (CPU/GPU and CUDA version). See PyTorch get started documentation for more details.

Instructions for M1/M2 MacBook users

You should use conda to set up an environment. If you are not familiar with conda, consult their get started documentation before you move forward.

Create conda environment from file

In your terminal run:

conda env create --file environment/conda_environment.yml

Activate an environment

conda activate ray-saturday

Test an environment

Check if Ray is installed correctly.

Start Python in the interactive mode

In your terminal run:

python3

Wait until you see >>> prompt.

Import ray

>>> import ray

Start Ray runtime locally

>>> ray.init()

If you see output like this, Ray is installed correctly:

2022-12-07 11:15:08,106 INFO worker.py:1519 -- Started a local Ray instance. View the dashboard at 127.0.0.1:8265

Feel free to open Ray dashboard at http://127.0.0.1:8265

What to do next?

Go to README to get started.

---

Troubleshooting

Check Ray installation documentation for more details and troubleshooting for issues with installing.