Policy output actiontuple

[andrewcoh]

Proposed change(s)

Useful links (Github issues, JIRA tickets, ML-Agents forum threads etc.)

Types of change(s)

• Bug fix
• New feature
• Code refactor
• Breaking change
• Documentation update
• Other (please describe)

Checklist

• Added tests that prove my fix is effective or that my feature works
• Updated the changelog (if applicable)
• Updated the documentation (if applicable)
• Updated the migration guide (if applicable)

Other comments

[chriselion]

How about keeping the continuous and discrete arguments, and adding e.g.

if continuous is not None:
  self.add_continuous(continuous)
# TODO same for discrete

?

[ervteng]

+1

[chriselion]

If this is identical to ActionTuple in functionality, but you want to make sure that they aren't accidentally used interchangeably, you could make them both derive from e.g. _ActionTupleBase

[chriselion]

I like this a lot better than the dictionary, although the class still feels clunky. I think you can keep the initializer args which should shrink the code a bit.

[ervteng]

If we add back the continuous and discrete parameters to ActionTuple, we can do something like:

Suggested change

	action_tuple.add_continuous(continuous)
	_continuous = continuous
	_discrete = np.column_stack() if self.discrete_size > 0 else None
	action_tuple = ActionTuple(_continuous, _discrete)