You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
There are two types of tests for agents: fast tests and slow tests. In fast tests, an agent is trained on an env for a small amount of steps. Slow tests continues until the agent solves the env.
The purpose of fast tests is to quickly check if code runs without error. However, it seems that in some agents' fast tests code that updates models is not executed due to too small amout of training steps, resulting in overlooked bugs like #399#400#402 . I think it is better to make sure that the update code is executed for each agent in its fast tests.
The text was updated successfully, but these errors were encountered:
There are two types of tests for agents: fast tests and slow tests. In fast tests, an agent is trained on an env for a small amount of steps. Slow tests continues until the agent solves the env.
The purpose of fast tests is to quickly check if code runs without error. However, it seems that in some agents' fast tests code that updates models is not executed due to too small amout of training steps, resulting in overlooked bugs like #399 #400 #402 . I think it is better to make sure that the update code is executed for each agent in its fast tests.
The text was updated successfully, but these errors were encountered: