steveKapturowski / tensorflow-rl

Implementations of deep RL papers and random experimentation
Apache License 2.0
177 stars 47 forks source link

Learning rate calculation of adam wrong? #20

Closed YurongYou closed 6 years ago

YurongYou commented 6 years ago

https://github.com/steveKapturowski/tensorflow-rl/blob/bcc9b2a966d7e5dee53f2a1fef1d5c355af85a79/algorithms/actor_learner.py#L319

steveKapturowski commented 6 years ago

Hi @YurongYou, I modified the code so I'm not changing the value of opt_st.lr.value. It appears to be functioning as expected as of commit 728f510a84681d49631e1a72294673e7950172ed

YurongYou commented 6 years ago

Great! I have to say your code is so well written! I learnt a lot from it. Thank you so much!

steveKapturowski commented 6 years ago

No problem! Glad you've found it useful