ML-Tracker is a project for creating an app that keeps track and progress of ML-Agents training sessions across your local machine and networked machines. ML-Agents is a powerful framework for creating intelligent agents in games and simulations using reinforcement learning, imitation learning, and other methods. ML-Tracker allows you to monitor and manage multiple ML-Agents training sessions and their outputs from a single interface.
- Monitor and manage multiple ML-Agents training sessions and their outputs from a single interface
- Connect and communicate with other ML-Trackers in a swarm or group or cluster
- Share and synchronize data and metrics among ML-Trackers
- Control and configure ML-Agents training sessions remotely
- Compare and evaluate ML-Agents results across different machines and environments
To use ML-Tracker, you need to have the following installed:
- [Microsoft Edge] browser
- [Unity Hub] and [Unity Editor] with ML-Agents package
- [Python] and [mlagents] package
- [ML-Explorer] project
To install ML-Tracker, you can clone this repository or download the zip file. Then, open the ML-Tracker folder in your preferred code editor.
To use ML-Tracker, you need to follow these steps:
- Launch the Microsoft Edge browser and open the ML-Tracker web page.
- Launch the Unity Hub and open the ML-Agents project that you want to track.
- Launch the Python terminal and run the mlagents-learn command with the appropriate configuration file and environment name.
- On the ML-Tracker web page, enter the name of your training session and the port number that you are using for communication between Unity and Python.
- Click on the "Start Tracking" button and watch the graphs and charts update in real time as your ML-Agents train.
- To connect with other ML-Trackers, click on the "Join Swarm" button and enter the IP address and port number of the ML-Tracker that you want to join.
- To share and synchronize data and metrics with other ML-Trackers, click on the "Sync Data" button and select the ML-Trackers that you want to sync with.
- To control and configure ML-Agents training sessions remotely, click on the "Remote Control" button and select the ML-Tracker that you want to control.
- To compare and evaluate ML-Agents results across different machines and environments, click on the "Compare Results" button and select the ML-Trackers that you want to compare with.
- To stop tracking, click on the "Stop Tracking" button and close the Unity and Python terminals.
ML-Tracker is licensed under the [MIT License].