robust_RL_multi_adversary

robust_RL_multi_adversary copied to clipboard

Hello! I want to have a better understanding of this code, so I need to get the distribution of action, such as the mean and standard deviation of Gaussian distribution....

c4cld

Added Bernoulli Bandit and appropriately updated run scripts.

kanaadp

The ma_crowd script fails for me with the folowing error: ``` File "/Users/kanaad/miniconda3/envs/sim2real/lib/python3.6/site-packages/ray/rllib/agents/trainer.py", line 415, in train raise e File "/Users/kanaad/miniconda3/envs/sim2real/lib/python3.6/site-packages/ray/rllib/agents/trainer.py", line 401, in train result = Trainable.train(self) File "/Users/kanaad/miniconda3/envs/sim2real/lib/python3.6/site-packages/ray/tune/trainable.py",...

kanaadp

eugenevinitsky

Because of the import of env_creator

kjang96

The RAM usage of the training goes up and up over time, leading to a failed experiment. Why?

eugenevinitsky

eugenevinitsky

The adversary applies actions at the latents of the autoencoder

eugenevinitsky

eugenevinitsky

Currently the rollout of a trained policy test generates a video but does not clear it up after.

eugenevinitsky