sconegym

sconegym is a gym environment for Reinforcement Learning (RL) using SCONE models.

To install sconegym, follow these steps:

1. Clone the sconegym repository from GitHub
2. Open a console and navigate to the local sconegym folder
3. Install dependencies using pip install -r requirements.txt
4. Install sconegym using pin install -e .

For more information on depRL, checkout the depRL project page.

To install depRL for use with sconegym, perform the following steps:

1. Make sure sconegym is installed
2. Clone the depRL repository from https://github.com/tgeijten/depRL/tree/sconegym
3. Open a console and navigate to the depRL folder
4. Install dependencies using pip install -r requirements.txt
5. Install depRL using pin install -e .

To test depRL with sconegym, perform these steps:

1. Navigate to depRL/experiments/hyfydy
2. Start an experiment using python -m deprl.main scone_run_h0918.yaml
3. Training starts and checkpoints are generated

To view the results, perform these steps:

1. Open SCONE Studio
2. Open the folder sconerun_h0918_v1 in the optimization results
3. Look for the checkpoints subfolder
4. Double-click on a checkpoint, or select multiple checkpoints and right-click and select Evaluate
5. The checkpoints are evaluated in the background; check the status bar to see the progress
6. When finished, look for folders starting with run_checkpoint
7. Double click on a .sto file to view and analyze the results