A test in `test_envs.py` launched a UnityEnvironment without mocking
the created communicator, leading to a port being reserved during the
test run. This in turn caused failures in later tests of
RpcCommunicator. This commit fixes that issue.
removed the Unity connect link as some folks were submitting issues there and getting lost. we do not actively monitor the connect channel and do not want to cause confusion
* Simplified rewards and observations; Determined better settings for training within a reasonable amount of time.
* Simplified Agent rewards; Added training section that discusses hyperparameters.
* Added note about DecisionFrequency.
* Updated screenshots and a small clarification in the text.
* Tested and updated using v0.6.
* Update a couple of images, minor text edit.
* Replace with more recent training stats.
* resolve a couple of minor review commnts.
* Increased the recommended batch and buffer size hyperparameter values.
* Fix 2 typos.
Some of our tests fail with the new pytest v4.0.x versions because of
a change to how fixtures work. Since these versions are the latest
release in PyPI now, unit tests will fail on a fresh install.
This change limits the version range in setup.py to fix the issue
until we can make a decision about upgrading.
* Check that worker port is available in RpcCommunicator
Previously the RpcCommunicator did not check the port or create the
RPC server until `initialize()` was called. Since "initialize"
requires the environment to be available, this means we might create
a new environment which connects to an existing RPC server running
in another process. This causes both training runs to fail.
As a remedy to this issue, this commit moves the server creation into
the RpcCommunicator constructor and adds an explicit socket binding
check to the requested port.
* Fixes suggested by Codacy
* Update rpc_communicator.py
* Addressing feedback: formatting & consistency
vincentpierre
GitHub
Jonathan Harper
Arthur Juliani
eshvk