From c04b2a395f82c149e9f3ad2adfdd04dd07794641 Mon Sep 17 00:00:00 2001 From: Shangtong Zhang Date: Tue, 10 Oct 2017 23:02:28 -0600 Subject: [PATCH] Update README --- README.md | 5 +---- 1 file changed, 1 insertion(+), 4 deletions(-) diff --git a/README.md b/README.md index 9444ee9..71876df 100644 --- a/README.md +++ b/README.md @@ -51,10 +51,7 @@ variance unbounded, which is also included in the implementation. ![Loading...](https://raw.githubusercontent.com/ShangtongZhang/DeepRL/master/images/DDPG-Pendulum-v0.png) -DDPG is extremely unstable and is the most difficult algorithm to tune from my experience. And it cannot solve -Continuous Lunar Lander or Bipedal Walker. I never see a public DDPG implementation without a fixed random seed - that can solve tasks other than the family of Pendulum. If you find a bug or some successful practice, it will - be much appreciated to let me know that. +Current DDPG implementation seems to have potential bugs, I'm now actively working on it. ## DPPO