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

Curriculum Learning | Academy Reset Parameters Resets to Serialized Values #3088

Closed afewvowels closed 4 years ago

afewvowels commented 4 years ago

Hello, I am absolutely having a blast with the ML Agents toolkit! It's so much fun building test chambers and watching AI find their way to the goal.

I've recently started using curriculum to change environment parameters as the learning process goes on and it's worked so far on my laptop computer. But I've got an environment set up on my desktop and every time the curriculum changes, the academy immediately resets with the original values I entered in manually. You can see these two lines of code from the command console which shows the curriculum has successfully registered a threshold and changed the academy reset parameters followed by a line resetting the reset parameters to the values that the academy has in its serialized fields in the inspector window.

INFO:mlagents.trainers:CubeJumper lesson changed. Now in lesson 5: wall_min_height -> 8.0, wall_max_height -> 11.0, platform_near_wall -> 1.0 INFO:mlagents.envs:Academy reset with parameters: wall_min_height -> 0.0, wall_max_height -> 2.0, platform_near_wall -> 0.0

Is there any way to prevent the reset parameters of the academy from resetting on curriculum changes?

Thanks!

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.