New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Series of safety environments #172

Merged
merged 10 commits into from Jun 12, 2016

Conversation

Projects
None yet
2 participants
@RafaelCosman
Copy link
Contributor

RafaelCosman commented Jun 12, 2016

This series has envs in three topics:

(1) off_switch (agents should not avoid being turned off)
(2) semi_supervised (agents should be able to learn from just occasionally sampling the reward)
(3) interpretability (agents should provide information about their plans)

TODO: Need to pass Nose tests.

super(OffSwitchCartpoleEnv, self).__init__()

def _render(self):
return super(OffSwitchCartpoleEnv, self)._render()

This comment has been minimized.

@gdb

gdb Jun 12, 2016

Collaborator

(No need to explicitly override these, FWIW.)

This comment has been minimized.

@RafaelCosman

RafaelCosman Jun 12, 2016

Contributor

Great point.

@RafaelCosman RafaelCosman force-pushed the RafaelCosman:safety-envs branch from c929249 to a855c01 Jun 12, 2016

@RafaelCosman RafaelCosman changed the title WIP: series of safety environments Series of safety environments Jun 12, 2016

@RafaelCosman RafaelCosman force-pushed the RafaelCosman:safety-envs branch 2 times, most recently from 8805f7e to 5abd16b Jun 12, 2016

@RafaelCosman RafaelCosman force-pushed the RafaelCosman:safety-envs branch from 5abd16b to 01662b1 Jun 12, 2016

@RafaelCosman

This comment has been minimized.

Copy link
Contributor

RafaelCosman commented Jun 12, 2016

@gdb @jietang I've squashed commits and rebased. This PR is ready for review.

env = InterpretabilityCartpoleActionsEnv()
env.reset()
for i in xrange(100):
print env.step([1, 0, 1, 0, 1, 0])

This comment has been minimized.

@gdb

gdb Jun 12, 2016

Collaborator

Can we make this into an automated test?

This comment has been minimized.

@RafaelCosman

RafaelCosman Jun 12, 2016

Contributor

Ok, I'll look into it.

@gdb

This comment has been minimized.

Copy link
Collaborator

gdb commented Jun 12, 2016

Super cool, looks good once real tests are added!

@RafaelCosman RafaelCosman force-pushed the RafaelCosman:safety-envs branch from ecae6e4 to e394564 Jun 12, 2016

@gdb

This comment has been minimized.

Copy link
Collaborator

gdb commented Jun 12, 2016

This looks good. I'm happy to fix up any test failures I see.

@gdb gdb merged commit c784b71 into openai:master Jun 12, 2016

1 check failed

continuous-integration/travis-ci/pr The Travis CI build failed
Details
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment