Hello!
I would like to ask if there is any documentation of SAC algorithm?
Thank you!
Hello!
I would like to ask if there is any documentation of SAC algorithm?
Thank you!
Hi,
The paper that the SAC trainer is based on can be found at [1801.01290] Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor
There’s also a blog post by the paper authors here that might be a gentler introduction: Soft Actor Critic—Deep Reinforcement Learning with Real-World Robots – The Berkeley Artificial Intelligence Research Blog
Ok !! Thank you very much!
I try to understand the source code of SAC that have been implemented in ML-Agents.
I try to understand the correspondence between the source code and the pseudocode of the article and I have been confused.
Can you give me some instructions of in which functions happens what according to the article.
If exists something that I can read and understand more things.
Thank you in advance!
I haven’t read through the whole paper or all of the code. We don’t have a more detailed documentation of the implementation anywhere. Is there a specific place that you’re wondering what the relation between the paper and the code is?
Thank you again very much! You have answered me very detailed in Clarification of PPO source code . My question for now is again in this post Clarification of PPO source code
→ If I will use the low-level Python API, I can build the environment from Unity and import it in the script.
At the end will a .nn file be created? So I can see my trained agent or this is not will happen?
Sorry, I didn’t see that message. I’ll reply in that thread.
Does not matter! It’s my fault for opening two topics.
Thank you very much for the clarifications