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.

Sign up for GitHub

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

Jump to bottom

Small improvements in the vtrace implementation #34

Merged

lespeholt merged 2 commits into google-deepmind:master from heiner:master

Mar 13, 2019

Contributor

heiner commented Mar 11, 2019

For x = np.random.randint(a, b), one has a <= x < b (unlike Python's standard random.randint(a,b) which is in [a, b], inclusive.
tf.scan has a reverse option is probably is marginally faster than inverting both its inputs and outputs.

Heiner added 2 commits

March 11, 2019 23:23


          Sample all actions in vtrace_test.

fcec1c2

Numpy's randint(a, b) samples from [a,b), including a but excluding b.


          Use tf.scan's reverse=True for vtrace's vs_minus_v_xs computation.

e331dc3

lespeholt merged commit 6c0c8a7 into google-deepmind:master

Collaborator

lespeholt commented Mar 13, 2019

Thanks!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment