search

Unity-Technologies / ml-agents

The Unity Machine Learning Agents Toolkit (ML-Agents) is an open-source project that enables games and simulations to serve as environments for training intelligent agents using deep reinforcement learning and imitation learning.
https://unity.com/products/machine-learning-agents
Other
16.93k stars 4.14k forks source link

Way forward for exposing Unity game for evolutionary algorithms that require no training #3139

Closed SurferZergy closed 4 years ago

SurferZergy commented 4 years ago

Hi, I would like to try evolution/genetic algorithm (https://arxiv.org/abs/1704.07075) which require no training a .nn model. My thought is to use gym as an environment and just perform the env.step based on calculating the reward using a forward model.

Anyways, it all works until i want to visualize what is happening, I think the steps are happening too fast, is there a way to slow it so i can see what is happening?

I know mlagents is probably treating it as a training process, but I believe these algorithms can be put directly into the brain without training.

ps. i'm on mlagents 0.10.1 right now, and using the gym-unity folder that came with that version. I can't do --slow (like https://github.com/Unity-Technologies/ml-agents/issues/833) since i'm not training. I don't think the train_model = true still applies.

surfnerd commented 4 years ago

Hi @zergy, You can change the training configuration in the Academy of the environment you are using by clicking on it in the scene view. The timeScale property could be set to 1 in order for you to run the simulation in real time instead of sped up time.

SurferZergy commented 4 years ago

thanks!

stale[bot] commented 4 years ago

This issue has been automatically marked as stale because it has not had activity in the last 14 days. It will be closed in the next 14 days if no further activity occurs. Thank you for your contributions.

github-actions[bot] commented 3 years ago

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.