hiro/ray
1
0
Fork 0
mirror of https://github.com/vale981/ray synced 2025-03-08 19:41:38 -05:00
Code Issues Projects Releases Packages Wiki Activity Actions
2335 commits 227 branches 67 tags 234 MiB
Commit graph

991 commits

Author SHA1 Message Date
Richard Liaw
ac792d70c8
[rllib] Add starcraft multiagent env as example (#3542) 2018-12-27 10:00:32 +08:00
Tianming Xu
b4f61dfd50 [rllib] Export policy model checkpoint (#3637) 
* Export policy model checkpoint

* update comment
 2018-12-27 08:43:06 +09:00
Richard Liaw
6e2d7a9ba1 [tune] Support Configuration Merging (#3584) 
* merge configs

* deep merge

* lint

* add resolve

* test
 2018-12-26 20:07:11 +09:00
Stan Wang
4ce3818be5 Average aggregated gradients before put in plasma store (#3631) 2018-12-26 20:03:11 +09:00
Yuhong Guo
1b98fb8238 Fix Jenkins test failures and function descriptor bug. (#3569) 
## What do these changes do?
1. Fix the Jenkins test failure by add driver id to Actor GCS Key.
2. Move `object_manager_test.py` from Jenkins to Travis.
 2018-12-25 23:31:44 -08:00
Robert Nishihara
5426234cd8 Update documentation to reflect 0.6.1 release. (#3622) 2018-12-24 11:10:04 -08:00
nam-cern
3d8f56409b Ensure numpy is at least 1.10.4 in setup.py (#2462) 
In the build script, numpy is specifically set at 1.10.4. We should also ensure that it is indeed the case in `setup.py`.
 2018-12-24 11:01:25 -08:00
Eric Liang
9f63119a83
[rllib] Allow development without needing to compile Ray (#3623) 
* wip

* lint

* wip

* wip

* rename

* wip

* Cleaner handling of cli prompt
 2018-12-24 18:08:23 +09:00
Devin Petersohn
c13b2685f5 [modin] Append to path to avoid namespace collision on development branches (#3621) 2018-12-23 23:58:56 -08:00
Alexey Tumanov
9b8d7573fe bump version from 0.6.0 to 0.6.1 (#3610) 2018-12-23 17:03:42 -08:00
Robert Nishihara
bb7ca3bae7 Upgrade flatbuffers version to 1.10.0. (#3559) 
* Upgrade flatbuffers version to 1.10.0.

* Temporarily change ray.utils.decode for backwards compatibility.
 2018-12-23 14:56:34 -08:00
Tianming Xu
deb26b954e [rllib] Export tensorflow model of policy graph (#3585) 
* Export tensorflow model of policy graph

* Add tests,examples,pydocs and infer extra signatures from existing methods

* Add example usage in export_policy_model comment

* Fix lint error

* Fix lint error

* Fix lint error
 2018-12-22 17:35:25 +09:00
Eric Liang
ddc97864df [rllib] Add requested clarifications to test requirement of contrib docs (#3589) 2018-12-21 11:02:02 -08:00
Richard Liaw
e046a5c767
[tune] resources_per_trial from trial_resources (#3580) 
Renaming variable due to user errors.
 2018-12-20 19:00:47 -08:00
Devin Petersohn
a174a46e02 Allowing multiple users to access the /tmp/ray file at the same time (#3591) 
* Allowing multiple users to access the /tmp/ray file at the same time

Previous sequence that caused this issue:
* User A starts ray with `ray.init` when /tmp/ray does not exist
* User B starts ray with `ray.init` and /tmp/ray now exists

User B will get a permissions error
Checking the permissions, /tmp/ray is 700

I have identified a race condition in `try_to_create_directory`
* Multiple processes try to create /tmp/ray at the same time
* chmod is either silently erroring or working properly within the race condition

Resolution: Move chmod outside of the check for whether the directory exists or not.

* Adding try except for users who do not own the directory
 2018-12-20 18:46:54 -08:00
Stephanie Wang
34bab6291c
Cleanup actor handle pickling code (#3560) 
* Cleanup actor handle pickling code

* remove unused

* fix

* lint
 2018-12-20 16:37:21 -08:00
Eric Liang
6bb1103930 [rllib] Avoid sample wastage with bad PPO configurations (#3552) 
## What do these changes do?

Previously we logged a warning if the PPO configuration would waste many samples. However, this didn't apply in the case of long episodes in `complete_episodes` batch mode, and also the amount of waste is up to 2x in common cases.

This pr:
- Estimates the number of sampling tasks needed to avoid over-sampling.
- Collects all sample results and never discards any. In principle this can degrade performance at large scale if certain machines are slower. Add a config flag to enable this legacy behavior.

## Related issue number

Closes: https://github.com/ray-project/ray/issues/3549
 2018-12-20 10:50:44 -08:00
Richard Liaw
ac48a58e4e
[tune] Reduce scope of variant generator (#3583) 
This PR provides a better error message when the generate_variants code
breaks. Also removes a comment about nesting dependencies.

This comes mainly as a hotfix solution for #3466. We should leave that issue open for future contribution 🙂
 2018-12-20 10:48:28 -08:00
Eric Liang
303883a3b6 [rllib] [rfc] add contrib module and guideline for merging (#3565) 
This adds guidelines for merging code into `rllib/contrib` vs `rllib/agents`. Also, clean up the agent import code to make registration easier.
 2018-12-20 10:44:34 -08:00
adoda
cf0c4745f4 [rllib] support running older version tensorflow(version < 1.5.0) (#3571) 2018-12-19 20:27:24 -08:00
Robert Nishihara
a5309bec7c Make README render properly on PyPI. (#3578) 
* Make README render properly in pypi.

* Add small logo

* temporary fix

* smaller image

* Remove image size.

* Add author and email to setup.py.
 2018-12-19 18:41:09 -08:00
Eric Liang
ffa6ee3ec8
[rllib] streaming minibatching for IMPALA (#3402) 
* mb impala

* fix

* paropt

* update

* cpu warn

* on cpu

* fix mb

* doc

* docs

* comment

* larger num

* early release

* remove grad clip

* only check loader count in multi gpu mode

* revert bad multigpu changes

* num sgd iter

* comment

* reuse optimizer

* add test

* par load test

* loosen test

* Update run_multi_node_tests.sh

* fix local mode

* Update agent.py
 2018-12-19 02:23:29 -08:00
Alexey Tumanov
c4cba98c75 Remove deprecation warnings when running actor tests (#3563) 
* remove deprecation warnings when running actor tests

* replacing logger.warn with logger.warning

* Update worker.py

* Update policy_client.py

* Update compression.py
 2018-12-18 17:04:51 -08:00
Yuhong Guo
fb33fa9097 Enable function_descriptor in backend to replace the function_id (#3028) 2018-12-18 18:53:59 -05:00
Yuhong Guo
75ddf7cca4 Fix 2 small bugs (#3573) 2018-12-18 14:52:21 -05:00
Eric Liang
db0dee573e
[rllib] Q-Mix implementation (Q-Mix, VDN, IQN, and Ape-X variants) (#3548) 2018-12-18 10:40:01 -08:00
opherlieber
854b06854f remove auto-concat of rollouts in AsyncSampler (#3556) 
* remove auto-concat of rollouts in AsyncSampler

* remote auto-concat test

* remove unused reference
 2018-12-17 13:54:52 -08:00
Robert Nishihara
417c7f2d6f Update arrow and remove plasma_manager references. (#3545) 2018-12-15 23:36:02 -08:00
Philipp Moritz
b3bf608608 Update arrow to reduce plasma IPCs. (#3497) 2018-12-14 23:49:37 -05:00
Richard Liaw
de3fdeb5b5
[autoscaler] Fix Error Handling for botocore (#3534) 
Unfortunately Boto generates error classes dynamically, so this catches
the expected error and raises the error if it is the wrong class.

Closes #3533.
 2018-12-14 00:20:49 -08:00
Hao Chen
e7b51cbd1b [xray] Implement Actor Reconstruction (#3332) 
* Implement Actor Reconstruction

* fix

* fix actor handle __del__

* fix lint

* add comment

* Remove actorCreationDummyObjectId

* address comments

* fix

* address comments

* avoid copy

* change log to debug

* fix error name
 2018-12-13 21:28:58 -08:00
Si-Yuan
84fae57ab5 Convert the raylet client (the code in local_scheduler_client.cc) to proper C++. (#3511) 
* refactoring

* fix bugs

* create client class

* create client class for java; bug fix

* remove legacy code

* improve code by using std::string, std::unique_ptr rename private fields and removing legacy code

* rename class

* improve naming

* fix

* rename files

* fix names

* change name

* change return types

* make a mutex private field

* fix comments

* fix bugs

* lint

* bug fix

* bug fix

* move too short functions into the header file

* Loose crash conditions for some APIs.

* Apply suggestions from code review

Co-Authored-By: suquark <suquark@gmail.com>

* format

* update

* rename python APIs

* fix java

* more fixes

* change types of cpython interface

* more fixes

* improve error processing

* improve error processing for java wrapper

* lint

* fix java

* make fields const

* use pointers for [out] parameters

* fix java & error msg

* fix resource leak, etc.
 2018-12-13 13:39:10 -08:00
Chunyang Wen
5dcc333199 [sgd] Modify: add interface for model (#3458) 
* Modify: add interface for model

* Modify: remove single quota and build; add metrics

* Modify: flatten into list of dict

* Update distributed_sgd.rst

* Modify: update format with scripts/format.sh

* Update sgd_worker.py
 2018-12-12 21:23:25 -08:00
Eric Liang
0e00533ed4
Different approach to removing RayGetError (#3471) 2018-12-12 20:30:51 -08:00
Eric Liang
32473cf22e
[rllib] Basic Offline Data IO API (#3473) 2018-12-12 13:57:48 -08:00
Richard Liaw
cc8f7db246
[docs] Improve cluster/docker docs (#3517) 
- Surfaces local cluster usage
 - Increases visability of these instructions
 - Removes some docker docs (that are really out of scope for Ray
 documentation IMO)

Closes #3517.
 2018-12-12 10:40:54 -08:00
Eric Liang
5f4a9cc713 [rllib] Rollout should preprocess observations; some cleanups (#3512) 
<!--
Thank you for your contribution!

Please review https://github.com/ray-project/ray/blob/master/CONTRIBUTING.rst before opening a pull request.
-->

## What do these changes do?

From https://groups.google.com/forum/#!topic/ray-dev/u-gybKK6-Ns
 2018-12-11 20:16:38 -08:00
Eric Liang
59f4743f20
[rllib] Run simple regressions tests for all algs in jenkins (#3498) 2018-12-11 17:21:53 -08:00
Richard Liaw
e0fbb68e47
[tune] Custom Logging, Trial Name (#3465) 
Adds support for custom loggers, custom trial strings, and custom sync commands. Closes #3034, #2985, and #3390.
 2018-12-11 13:41:59 -08:00
Eric Liang
52df4dfc6f
[rllib] Fix multiagent_two_trainer test (#3509) 
* update

* fix

* dict ordre

* fix

* fix
 2018-12-11 00:16:39 -08:00
Richard Liaw
1f4a01cff6 [tune] Fix PyTorch example after PyTorch v1 (#3500) 
* [tune]

* fix

* lint

* fix
 2018-12-10 12:00:53 -08:00
Eric Liang
962f18756b [autoscaler] Use fixed timestamp to check against health timeouts (#3503) 2018-12-10 14:58:27 -05:00
Eric Liang
ce388a45cf
[rllib] Learner should not see clipped actions (#3496) 2018-12-09 21:57:11 -08:00
Philipp Moritz
87c0d24579
[sgd] Add file lock to protect compilation of sgd op (#3486) 
* add file lock to protect compilation of sgd op

* lint

* update

* fix

* fix

* lint

* update

* rebase on arrow

* Update sgd_worker.py
 2018-12-09 13:52:40 -08:00
Eric Liang
cffe8f9806 Add option to evict keys LRU from the sharded redis tables (#3499) 
* wip

* wip

* format

* wip

* note

* lint

* fix

* flag

* typo

* raise timeout

* fix

* optional get

* fix flag

* increase timeout in test

* update docs

* format
 2018-12-09 05:48:52 -08:00
Eric Liang
7aec357501
[rllib] Multi-GPU support for Multi-Agent PPO (#3479) 
* wip

* fix

* remove check

* fix null

* revert

* lint and kl

* also fix rollout
 2018-12-08 18:02:33 -08:00
Eric Liang
8b5827b9da
[rllib] Better document which methods are abstract and which ones are overrides (#3480) 2018-12-08 16:28:58 -08:00
Eric Liang
462e6ef066
[rllib] Use smoothed version of collect metrics for DQN (#3491) 
* fix

* lint
 2018-12-07 18:36:23 -08:00
Tianming Xu
f6490f9bef Resolve no handlers could be found for logger 'ray.worker' when importing ray (#3483) 2018-12-06 20:46:53 -08:00
Eric Liang
8395523f81
[rllib] Copy data before passing to Ape-X learner thread (fixes transient plasma crashes) (#3484) 2018-12-06 18:01:11 -08:00