This repository has been archived by the owner on Mar 11, 2021. It is now read-only.

Remove tf.control_dependencies from dual_net.py #870

Open

tommadams opened this issue Aug 16, 2019 · 0 comments

Contributor

tommadams commented Aug 16, 2019

In order to get multiple models running on the same TPU, we had to add the following hack to work around a bug in TensorFlow:

minigo/dual_net.py

Line 523 in d5e9697

epoch_time = tf.constant(t, name='epoch_time_%d' % t)

This bug got fixed in TF 1.12.1, so we can remove our hack now:
tensorflow/tensorflow@735d26a#diff-c344ded643ece6ebf6370acab3a9a524L306

The text was updated successfully, but these errors were encountered:

sethtroisi mentioned this issue

Revert TPU hack #963

Closed

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