Commit Graph

448 Commits

Author SHA1 Message Date
Trevor Blackwell
17bde31080 Merge branch 'master' into trace_recording 2016-09-07 19:41:46 -07:00
Trevor Blackwell
0f02e3a817 Change trace recording from a monitor to a wrapper. 2016-09-07 19:38:24 -07:00
Greg Brockman
4e88d00e5e Bump Atari timestep limits by 10x (#330) 2016-09-05 23:39:32 -07:00
Greg Brockman
e77c3095d6 Bump version 2016-09-05 16:02:53 -07:00
Greg Brockman
0278f273fa Revert wrappers to having their own monitor
Otherwise you end up with multiple calls to the same monitor at each
step.
2016-09-05 16:00:31 -07:00
Greg Brockman
f5025d329c Bump version 2016-09-05 10:31:00 -07:00
Greg Brockman
a6ec54415c By default, Wrappers shouldn't have their own monitors 2016-09-05 10:30:54 -07:00
Greg Brockman
9b8472726c Check for hasattr _monitor before invoking monitor 2016-09-04 19:41:11 -07:00
Greg Brockman
2cd674aaa8 Bump version 2016-09-04 01:44:39 -07:00
Greg Brockman
53eca9c55c Add optional tags 2016-09-04 01:44:27 -07:00
Greg Brockman
6c6bd391a5 Add non-underscore methods for Wrappers 2016-09-04 01:44:20 -07:00
Greg Brockman
2c580d94d7 Bump version 2016-09-04 00:39:15 -07:00
Greg Brockman
f30ff469b8 [WIP] Start adding Filter API (#329)
Expand Wrapper API
2016-09-04 00:38:03 -07:00
Tom Brown
1452dc3ca2 Bump to 0.2.7 2016-09-02 18:51:20 -07:00
Tom Brown
cf028fc0e3 Add AtariNoFrameskip-v0 (#328)
* Add PongDeterministicNoFrameskip-v0

* Update __init__.py

* Update __init__.py
2016-09-02 17:26:33 -07:00
Trevor Blackwell
9b882e7c7b Recording traces
With monitor is started with `.start(…save_trace=True)`, record
complete action/observation/reward tuples for any environment. Does it
efficiently in picked numpy arrays in the same directly as the other
monitoring files.
2016-09-02 11:22:37 -07:00
Trevor Blackwell
ec3f2e7f3e Formatting 2016-09-02 10:27:54 -07:00
Trevor Blackwell
4846d02288 Explain how to create new environments in a separate repo 2016-09-02 10:26:47 -07:00
Daniel Pepper
113abbfce2 limit the action on mountain car so that there isn't a trivial win scenario. otherwise, you can simply pass action=10 into the env.step() and win. see https://gym.openai.com/evaluations/eval_4oWGpaKrRLKssArKVss0sQ for example of why this limit is needed. (#303)
also did a couple minor cleanups while I was in there
2016-08-30 16:31:34 -07:00
Dr. Kashif Rasul
ebb2f29273 added ffmpeg to dockerfile and fixed test (#325)
* added ffmpeg to dockerfile and fixed test

* use ubuntu media repo for ffmpeg

* ffmpeg on travis 🍰

* install libav instead
2016-08-29 16:12:29 -07:00
John Schulman
a84edbc78b avoid opening window on close 2016-08-25 17:19:54 -07:00
Greg Brockman
51b62ef116 Bump version 2016-08-25 08:58:23 -07:00
Greg Brockman
e3e735b1ad Add Atari environments with deterministic frameskip 2016-08-25 08:58:09 -07:00
Greg Brockman
d3df8dcd05 Bump version 2016-08-24 22:24:05 -07:00
catherio
fd8101a65d Move pygame scoreboard registration to scoreboard-gym (#320)
Move pygame scoreboard registration
2016-08-24 16:43:52 -07:00
Rafael Cosman
934baaf900 Fixes bug in comment (#262) 2016-08-24 16:09:04 -07:00
Oleg Klimov
ee2c0243c0 LunarLanderContinuous (#307)
* New LunarLanderContinuous, LunarLander-v2 remains exactly the same, no version bump.

* keyboard_agent.py works again.
2016-08-24 16:08:32 -07:00
Olivier Sigaud
c97551e8e5 added continuous mountain car v0 (#306)
* added continuous mountain car v0

* spotted that the action should be a vector, not a scalar

* fixed bug on action format

* bug fixed in scoreboard registering

* the observation (aka state) should be a numpy array

* added comment on reward range as requested
2016-08-24 14:10:58 -07:00
Philip Paquette
4b9984c00e Moved doom to gym_doom (#319) 2016-08-24 08:24:10 -07:00
rafal
3d29fb541b bump version 2016-08-24 00:18:11 -07:00
Rafal Jozefowicz
c11cacd50d Remove unnecessary copies from atari environments (#317) 2016-08-24 00:15:27 -07:00
Jesse Cooper
932ed300f4 Fix error when running min on empty list (#314)
* Fix error when running min on empty list

Running min on an empty list (which occurs on first episode from `if len(content['timestamps'])==0: continue`) causes it to throw an error

* Update monitor.py

Return 0 for initial_reset_timestamp when no initial_reset_timestamps
2016-08-23 22:55:42 -07:00
catherio
7f2a52bf19 Create separate descritpions for PyGame environments, using text copied from the PLE website (#313)
Create separate descriptions for PyGame environments
2016-08-23 20:01:43 -07:00
Luis Sobrecueva
105f0365fa Added PLE env to scoreboard (#299) 2016-08-23 18:02:09 -07:00
catherio
295812fd40 Allow envs to be registered on the scoreboard but not in gym core (#311) 2016-08-23 17:00:47 -07:00
John Schulman
dbf7e51023 define actual methods for register, make, spec 2016-08-20 16:05:50 -07:00
Greg Brockman
b35b24e69d Bump version 2016-08-18 20:02:21 -07:00
Greg Brockman
fd051e0895 Drop 'ending episode' message to debug 2016-08-18 20:02:04 -07:00
John Schulman
75b6d9993f minor fix 2016-08-18 09:26:18 -07:00
Greg Brockman
e2c9e84a51 Bump version 2016-08-17 15:55:14 -07:00
Greg Brockman
ee02589eec Don't have envs be aware of wrappers (#296)
You can always achive the same effect by exposing a wrapping function:

def MyEnv():
  return MyWrapper(MyUnwrappedEnv())
2016-08-17 15:16:45 -07:00
Greg Brockman
b3ef84839e Add training directory to logging 2016-08-17 14:00:53 -07:00
Greg Brockman
40fc7d57ce Have env.close also close the monitor and any open windows 2016-08-17 13:59:35 -07:00
John Schulman
c6a3f66537 log error in load_results 2016-08-17 12:51:29 -07:00
Greg Brockman
a74fcb5dd1 Bump version 2016-08-17 12:00:16 -07:00
Greg Brockman
8a9bcdf061 Don't write stats file unless requested either 2016-08-17 11:59:29 -07:00
Greg Brockman
cfd90f38c8 Bump version 2016-08-17 11:26:55 -07:00
Greg Brockman
922fc56d2f Improve performance for short episodes by only writing upon reset when requested 2016-08-17 11:26:55 -07:00
Philip Paquette
14f8c9a3f2 Wrapper - Common frame skipping (#293)
* Common frame skipping

* Common frame skipping - Fixes

* Raising error if skip.stepcount already in info.
2016-08-14 19:24:26 -07:00
Philip Paquette
cc1483deb1 Renamed HighLow to MultiDiscrete, added adapters and better documentation (#292) 2016-08-14 13:18:28 -07:00