Commit Graph

9 Commits

Author SHA1 Message Date
Philip Paquette
aff7a643cc Doom - Same Action Space Across Environments (#157)
* Doom - Added reward_threshold and timestep_limit for all environments

* Doom - Returning all available game variables

* Doom - Moved _seed to doom_env to avoid repetition in every environment

* Doom - Added ALT_ATTACK and made all action_space equivalent (same controls between environments).

* Doom - Actions can either be a short list of allowed actions or the full list of 41 commands

* Doom - Returning black observation space on error or is_finished, rather than empty list (which was triggering an error)

* Doom - HighLow.sample() returns the small list.

* Doom - Updated difficulty for some missions

* Doom - Fixed inconsistency between controls.md and deathmatch.cfg

* Doom - Issue #168 - Remove sleep statement from DoomEnv render

* Doom - Only using full action space (43 keys)

- Added 'normal', 'fast' and 'human' mode
- Set non-deterministic to True
- Set video.frames_per_second to 35
- Properly returning game variables

* Replaced warnings.warn by logger.warn

* Doom - Added NUM_ACTIONS and action_idx instead of x

* Doom - Added NUM_ACTIONS and action_idx instead of x

* Doom - reset() only calls game.new_episode() after first call

* Doom is now deterministic

* Doom - Partial fix for issue #167 - DoomDeathmatch environment crashes sporadically

* Doom - Standardized envs, simplified _reset

* Doom - Removed temporary fix for issue #167

* Doom - Added scoreboard summary and description
2016-06-14 15:57:47 -07:00
Rafael Cosman
f254dd197e adds example usage to spaces (#173) 2016-06-11 23:10:58 -07:00
Greg Brockman
8a535ca6f2 Switch to a global PRNG for action/observation spaces (#144)
cf 58e6aa95e5 (commitcomment-17669277)
2016-05-30 18:07:59 -07:00
Greg Brockman
58e6aa95e5 [WIP] add support for seeding environments (#135)
* Make environments seedable

* Fix monitor bugs

- Set monitor_id before setting the infix. This was a bug that would yield incorrect results with multiple monitors.
- Remove extra pid from stats recorder filename. This should be purely cosmetic.

* Start uploading seeds in episode_batch

* Fix _bigint_from_bytes for python3

* Set seed explicitly in random_agent

* Pass through seed argument

* Also pass through random state to spaces

* Pass random state into the observation/action spaces

* Make all _seed methods return the list of used seeds

* Switch over to np.random where possible

* Start hashing seeds, and also seed doom engine

* Fixup seeding determinism in many cases

* Seed before loading the ROM

* Make seeding more Python3 friendly

* Make the MuJoCo skipping a bit more forgiving

* Remove debugging PDB calls

* Make setInt argument into raw bytes

* Validate and upload seeds

* Skip box2d

* Make seeds smaller, and change representation of seeds in upload

* Handle long seeds

* Fix RandomAgent example to be deterministic

* Handle integer types correctly in Python2 and Python3

* Try caching pip

* Try adding swap

* Add df and free calls

* Bump swap

* Bump swap size

* Try setting overcommit

* Try other sysctls

* Try fixing overcommit

* Try just setting overcommit_memory=1

* Add explanatory comment

* Add what's new section to readme

* BUG: Mark ElevatorAction-ram-v0 as non-deterministic for now

* Document seed

* Move nondetermistic check into spec
2016-05-29 09:07:09 -07:00
Greg Brockman
a20719c8e3 Revert "Feature/len (#68)"
This reverts commit 705d024993.
2016-05-09 18:34:44 -07:00
moto
705d024993 Feature/len (#68)
* Add __len__ to discrete

* Add __len__ to tuple
2016-05-09 17:59:55 -07:00
Jonas Schneider
618c7fda5e Don't break when passing strange values into DiscreteSpace#contains 2016-05-03 22:27:06 -04:00
Jonas Schneider
1fd5af8b4e Apparently Python3 disallows casting numpy scalars into ints. Work around that. 2016-04-27 18:31:32 -07:00
Greg Brockman
e8f2980603 Initial release. Hello world :). 2016-04-27 08:00:58 -07:00