search

araffin / learning-to-drive-in-5-minutes

Implementation of reinforcement learning approach to make a car learn to drive smoothly in minutes
https://towardsdatascience.com/learning-to-drive-smoothly-in-minutes-450a7cdb35f4
MIT License
284 stars 88 forks source link

learning_rate remains unchange in custom_sac.py #11

Closed wangyubin112 closed 4 years ago

wangyubin112 commented 4 years ago

I notice that constfn() in stable_baselines.ppo2.ppo2 returns a constant value, thus self.learning_rate() remains constant. Based on the src_code, I guess it is not what you expect, right?

araffin commented 4 years ago

Hello, This is completely expected behavior. See code here for linear/constant schedule:https://github.com/araffin/learning-to-drive-in-5-minutes/blob/89a3b2ca040014cb2193ad3fe88636de146f49ce/train.py#L88-L98

wangyubin112 commented 4 years ago

Hi, OK, thank you for the remind!