74 Policy evaluation and training cli (rllib) #85
Conversation
| @@ -192,8 +192,6 @@ def __init__(self, | |||
| self.num_resets = 0 | |||
| self.distance_map = DistanceMap(self.agents, self.height, self.width) | |||
|
|
|||
| self.action_space = [5] | |||
There was a problem hiding this comment.
Is this removal safe? Remove from flatland.core.Environment.env as well or redefine there?
There was a problem hiding this comment.
What does mean removal safe - remove from ... ? You like to delete the file or just action_space ?
flatland/ml/observations/flatten_tree_observation_for_rail_env.py
Outdated
Show resolved
Hide resolved
chenkins
changed the title
chenkins
force-pushed
the
74-policy-evaluation-and-training-cli
branch
from
daec963 to
3f903ca
Compare
flatland/ml/observations/flatten_tree_observation_for_rail_env.py
Outdated
Show resolved
Hide resolved
| self.env_renderer = RenderTool(wrap) | ||
|
|
||
| self.action_space: gym.spaces.Dict = spaces.Dict({ | ||
| # TODO document why str is necessary - is it? |
There was a problem hiding this comment.
TODO document why str is necessary - is it?
There was a problem hiding this comment.
str(i) === i don't understand this !!!
Do you like to get for each agent a Discrete(5) : |agents| x |Actions space|
| # TODO dqn not working: rewards = scipy.signal.lfilter([1], [1, -gamma], raw_rewards[::-1], axis=0)[ | ||
| # ~~~~~~~~~~~^^^^^^ | ||
| # TypeError: unhashable type: 'slice' | ||
| # "DQN", |
There was a problem hiding this comment.
TODO dqn not working
Co-authored-by: Serge Croisé <[email protected]>
|
|
||
|
|
||
| # defaults from Flatland 3 Round 2 Test_0, see https://flatland.aicrowd.com/challenges/flatland3/envconfig.html | ||
| def env_creator(n_agents=7, |
There was a problem hiding this comment.
maybe use env_generator to use same naming as for rail and line generators?
flatland/ml/ray/wrappers.py
Outdated
| return RayMultiAgentWrapper(wrap, render_mode) | ||
|
|
||
|
|
||
| def ray_env_creator(render_mode: Optional[str] = None, **kwargs) -> RayMultiAgentWrapper: |
There was a problem hiding this comment.
same questions as above regarding naming _creator or _generator
| @@ -192,8 +192,6 @@ def __init__(self, | |||
| self.num_resets = 0 | |||
| self.distance_map = DistanceMap(self.agents, self.height, self.width) | |||
|
|
|||
| self.action_space = [5] | |||
There was a problem hiding this comment.
What does mean removal safe - remove from ... ? You like to delete the file or just action_space ?
| self.env_renderer = RenderTool(wrap) | ||
|
|
||
| self.action_space: gym.spaces.Dict = spaces.Dict({ | ||
| # TODO document why str is necessary - is it? |
There was a problem hiding this comment.
str(i) === i don't understand this !!!
Do you like to get for each agent a Discrete(5) : |agents| x |Actions space|
Changes
pyproject.toml, newrequirements-ml.txtand slimmerrequirements[-dev].txt, keep coregymfree.flatland.mlmoduletests.ml(tests becomes a Python module)flatland.contribs)Related issues
Closes #74.
Fixes #23.
Closes #75
Checklist
docsfolder for relevant behavior changes. If you made important user-facingchanges, describe them under the
[Unreleased]tag inCHANGELOG.md.pyproject.tomlfile.Requirement files have been updated by running
tox -e requirements.required to run successfully.
CONTRIBUTING.mdare followed.