bsuite
bsuite copied to clipboard
google-deepmind
bsuite is a collection of carefully-designed experiments that investigate core capabilities of a reinforcement learning (RL) agent
Without this fix, this tile: ``` #@title overall score as radar plot (double-click to show/hide code) BSUITE_SCORE = summary_analysis.bsuite_score(DF, SWEEP_VARS) BSUITE_SUMMARY = summary_analysis.ave_score_by_tag(BSUITE_SCORE, SWEEP_VARS) BSUITE_SUMMARY __radar_fig__ = summary_analysis.bsuite_radar_plot(BSUITE_SUMMARY, SWEEP_VARS) ```...
Hi @iosband and @yotam, Hope you're doing well! Following our discussions, I added MDP Playground experiments into bsuite for the following dimensions of MDP Playground: **Delay**, **Transition Noise**, **Reward Noise**,...
Hi there, I am using `bsuite` to implement the dqn in _Human level control through deep reinforcement learning_. Whilst using the DMEnv wrapper for gym environments, I found useful for...
1. The description of stochastic sea environment says "adds N(0,1) noise to the end of states of the chain", but in line 125, noisy reward were only added when "column"...
Hi, I believe the default value for the number of ensemble should be more than 1. Otherwise the default agent does not solve Deep Sea. I set it to 20...
Current implementation thrashes the cpu ---> extremely slow running the sweep. Suggested fix here, but untested --- I get other errors trying to run the default agents: 1) transformed.apply expecting...
just a boolean flag to turn off the loading message when a bsuite env is created (in case it interferes with other messages)
Hello Ian and others! I'm having a look at `bsuite` after Ian Osband's talk at the Simons Institute Deep RL workshop. After spending a few minutes browsing the documentation and...
Metadata
Owner
Metadata
bsuite is a collection of carefully-designed experiments that investigate core capabilities of a reinforcement learning (RL) agent