search

HubFire / Muti-branch-DDPG-CARLA

A tensorflow implemention of ECCV2018 Paper：CIRL: Controllable Imitative Reinforcement Learning for Vision-based Self-driving
78 stars 21 forks source link

How many episodes or hours did you use to train your model? #2

Open anotherkey opened 5 years ago

anotherkey commented 5 years ago

I have been training it for about 4000 episodes(about 7 hours on one NVIDIA GeForce GTX1060 GPU), but the performance is still not good.I don't know how to observe the training process such as loss with the help of tensorboard. Could you please give me some advice?

merged = tf.summary.merge_all() sess.run(merged,feed_dict={})

1.Where should I run merged? 2.What is the content of the feed_dict?

I believe your paper and code are both excellent.What I am worried about is whether rl algorithms can work as well as the paper.

gnniaaa commented 3 years ago

I wonder how to train, maybe just run mai.py? But it doesn't work for me.