Docs changes for the Side Channels feature by vincentpierre · Pull Request #3011 · Unity-Technologies/ml-agents

vincentpierre · 2019-12-02T20:01:34Z

No description provided.

chriselion · 2019-12-02T20:16:03Z

 '--run-id': 'first-run',
 '--save-freq': '50000',
 '--seed': '-1',
- '--slow': False,


Actually, none of this is accurate anymore. It looks like

INFO:mlagents.trainers:CommandLineOptions(debug=False, num_runs=1, seed=-1, env_path=None, run_id='first-run', load_model=False, train_model=True, save_freq=50000, keep_checkpoints=5, base_port=5005, num_envs=1, curriculum_folder=None, lesson=0, slow=False, no_graphics=False, multi_gpu=False, trainer_config_path='config/trainer_config.yaml', sampler_file_path=None, docker_target_name=None, env_args=None, cpu=False) INFO:mlagents.envs:Start training by pressing the Play button in the Unity Editor.

today, Maybe just delete this part?

I agree to delete it. It does not add anything.

Co-Authored-By: Chris Elion <chris.elion@unity3d.com>

…Technologies/ml-agents into develop-side-channel-docs

Co-Authored-By: Chris Elion <chris.elion@unity3d.com>

chriselion · 2019-12-02T21:59:26Z

+* `CustomResetParameters` are now removed.
+* `reset()` on the Low-Level Python API no longer takes a `train_mode` argument. To modify the performance/speed of the engine, you must use an `EngineConfigurationChannel`
+* `reset()` on the Low-Level Python API no longer takes a `config` argument. `UnityEnvironment` no longer has a `reset_parameters` field. To modify float properties in the environment, you must use a `FloatPropertiesChannel`
+* The Academy no longer has a `Training Configuration` nor `Inference Configuration` field in the inspector. To modify the configuration form the Low-Level Python API, use an `EngineConfigurationChannel`. To modify it during training, use the new command line arguments `--width`, `--heigth`, `--quality-level`, `--time-scale` and `--target-frame-rate` in `mlagents-learn`.


Maybe something like "These default to the previous default values of the Training Configuration fields. If you weren't using --slow, you don't need to change anything. If you were running training with --slow, you should now pass --quality-level=... etc" (it might be better in a "Steps to Migrate" section" with slightly different wording)

I added :

Steps to Migrate

If you had a custom Training Configuration in the Academy inspector, you will need to pass your custom configuration at every training run using the new command line arguments --width, --heigth, --quality-level, --time-scale and --target-frame-rate.

If you were using --slow in mlagents-learn, you will need to pass your old Inference Configuration of the Academy inspector with the new command line arguments --width, --heigth, --quality-level, --time-scale and --target-frame-rate instead.

chriselion

One minor comment on the migration guide, looks good otherwise.

ervteng · 2019-12-02T22:17:15Z

+* `CustomResetParameters` are now removed.
+* `reset()` on the Low-Level Python API no longer takes a `train_mode` argument. To modify the performance/speed of the engine, you must use an `EngineConfigurationChannel`
+* `reset()` on the Low-Level Python API no longer takes a `config` argument. `UnityEnvironment` no longer has a `reset_parameters` field. To modify float properties in the environment, you must use a `FloatPropertiesChannel`
+* The Academy no longer has a `Training Configuration` nor `Inference Configuration` field in the inspector. To modify the configuration form the Low-Level Python API, use an `EngineConfigurationChannel`. To modify it during training, use the new command line arguments `--width`, `--heigth`, `--quality-level`, `--time-scale` and `--target-frame-rate` in `mlagents-learn`.


Spelling on --height

Also isn't this waaay too many options to do command line? I'd assume for most things you'd want to change all of them.

For us, the defaults are pretty standard. Width and height are not really useful. Quality settings do not make a huge difference and can be left at default value. Target frame rate of -1 is default and should not change. The only one that really changes is time scale

Good catch on the typo

Sounds good. So in training we'd just have them pass --time-scale=whatevertheywant. This is actually pretty nice, rather than baking it into the game.

ervteng · 2019-12-02T22:17:39Z

+### Important changes
+* `CustomResetParameters` are now removed.
+* `reset()` on the Low-Level Python API no longer takes a `train_mode` argument. To modify the performance/speed of the engine, you must use an `EngineConfigurationChannel`
+* `reset()` on the Low-Level Python API no longer takes a `config` argument. `UnityEnvironment` no longer has a `reset_parameters` field. To modify float properties in the environment, you must use a `FloatPropertiesChannel`


Would link to the Python API page here

ervteng · 2019-12-02T22:20:36Z

+__Note__ : A side channel will only send/receive messages when `env.step` is
+called.
+
+#### EngineConfigurationChannel


Would add a two-line example for creating each of these, and how to pass the list into the UnityEnvironment. Unless that's somewhere else in the docs?

okay will do

ervteng · 2019-12-02T22:50:35Z


-To enable variations in the environments, we implemented `Reset Parameters`. We
+To enable variations in the environments, we implemented `Reset Parameters`.
+`ResetParameters` are `Academy.FloatProperties` that are used only when


Space in Reset Parameters.

So what differentiates these two is that FloatProperties could (technically) be used anytime but our example envs only use them at reset right? The phrasing is a bit confusing but I think that makes sense.

ervteng · 2019-12-02T22:55:28Z

 To enable variations in the environments, we implemented `Reset Parameters`.
-`ResetParameters` are `Academy.FloatProperties` that are used only when
+`Reset Parameters` are `Academy.FloatProperties` that are used only when
 reseting the environment. We


Also spelling on the word reseting -> resetting

ervteng

had one last spelling comment but other than that LGTM

* [WIP] Side Channel initial layout * Working prototype for raw bytes * fixing format mistake * Added some errors and some unit tests in C# * Added the side channel for the Engine Configuration. (#2958) * Added the side channel for the Engine Configuration. Note that this change does not require modifying a lot of files : - Adding a sender in Python - Adding a receiver in C# - subscribe the receiver to the communicator (here is a one liner in the Academy) - Add the side channel to the Python UnityEnvironment (not represented here) Adding the side channel to the environment would look like such : ```python from mlagents.envs.environment import UnityEnvironment from mlagents.envs.side_channel.raw_bytes_channel import RawBytesChannel from mlagents.envs.side_channel.engine_configuration_channel import EngineConfigurationChannel channel0 = RawBytesChannel() channel1 = EngineConfigurationChannel() env = UnityEnvironment(base_port = 5004, side_channels = [channel0, channel1]) ``` * renamings * addressing comments * Logging a message when an unknown side channel number has been received by Unity * Addressing comments * renamings * renamings * Adding FloatProperties to the side channels (#2968) * renaming m_SideChannelsDict to m_SideChannel * renaming and some comments * renaming and adding a GetAndClearReceivedMessages() in the RawBytesSideChannel * micro-optimization * more errors and some nit * addressing comments * Using little-endian format in Python * adding some comments * Code comments * some changes and added the unit tests on both Python and C# * removing default default in get default * Update UnitySDK/Assets/ML-Agents/Scripts/SideChannel/SideChannel.cs Co-Authored-By: Chris Elion <chris.elion@unity3d.com> * Update ml-agents-envs/mlagents/envs/side_channel/raw_bytes_channel.py Co-Authored-By: Chris Elion <chris.elion@unity3d.com> * addressing comments * fixing tests * removing the arguments to reset and the property reset_parameters on the UnityEnvironment * curriculum works but removed the check for reset parameters in the scene * processing side channels before the reset command * Removing engine configuration from C# * Engine configuration removed * fixing the tests * Update ml-agents-envs/mlagents/envs/subprocess_env_manager.py Co-Authored-By: Chris Elion <chris.elion@unity3d.com> * Removing register callbacks with empty calls in FloarProperties * Clamp instead of min max * removing the brain names from the environment.py print * renaming reset_parameters to get properties * made a default engine config * bug fix * Empty commit * Docs changes for the Side Channels feature (#3011) * Docs changes for the Side Channels feature * replace deprecated with removed on the CustomResetPratmeters` * Update docs/Python-API.md Co-Authored-By: Chris Elion <chris.elion@unity3d.com> * Update docs/Training-Generalized-Reinforcement-Learning-Agents.md Co-Authored-By: Chris Elion <chris.elion@unity3d.com> * Removing the console outputs in the docs * Update docs/Training-ML-Agents.md Co-Authored-By: Chris Elion <chris.elion@unity3d.com> * replace does not work with ignored * adding a note on side channels * adding some steps to migrate * addressing comments * adding more docs to the LL-API * added a blob on how to access the properties in C# * adding space between ResetParameters * fix typo * bug fixes * addressing comments

MarkTension · 2019-12-09T09:06:17Z

Heya, I was directed here from this feature request. Where exactly can I find the part that covers the tensorboard C# variable functionality? Will my input still be relevant?

Docs changes for the Side Channels feature

aa62543

vincentpierre requested a review from chriselion December 2, 2019 20:01

vincentpierre self-assigned this Dec 2, 2019

replace deprecated with removed on the CustomResetPratmeters`

3d05e45

chriselion reviewed Dec 2, 2019

View reviewed changes

Comment thread docs/Python-API.md Outdated

chriselion reviewed Dec 2, 2019

View reviewed changes

Comment thread docs/Training-Generalized-Reinforcement-Learning-Agents.md Outdated

chriselion reviewed Dec 2, 2019

View reviewed changes

Comment thread docs/Training-ML-Agents.md Outdated

vincentpierre and others added 6 commits December 2, 2019 12:20

Update docs/Python-API.md

dcef7ad

Co-Authored-By: Chris Elion <chris.elion@unity3d.com>

Update docs/Training-Generalized-Reinforcement-Learning-Agents.md

0a5ac05

Co-Authored-By: Chris Elion <chris.elion@unity3d.com>

Removing the console outputs in the docs

68133eb

Merge branch 'develop-side-channel-docs' of https://github.com/Unity-…

7a2db2f

…Technologies/ml-agents into develop-side-channel-docs

Update docs/Training-ML-Agents.md

efe256e

Co-Authored-By: Chris Elion <chris.elion@unity3d.com>

replace does not work with ignored

dbc6d09

vincentpierre requested review from chriselion and ervteng December 2, 2019 21:03

vincentpierre added 2 commits December 2, 2019 13:29

adding a note on side channels

cff357b

Merge branch 'develop-side-channel-2' into develop-side-channel-docs

098653f

chriselion reviewed Dec 2, 2019

View reviewed changes

chriselion approved these changes Dec 2, 2019

View reviewed changes

shihzy mentioned this pull request Dec 2, 2019

Assigning custom c# variables to summary in tensorboard #2984

Closed

adding some steps to migrate

b754d01

ervteng reviewed Dec 2, 2019

View reviewed changes

vincentpierre added 2 commits December 2, 2019 14:22

addressing comments

8cca0de

adding more docs to the LL-API

2281a91

ervteng reviewed Dec 2, 2019

View reviewed changes

vincentpierre added 2 commits December 2, 2019 14:53

added a blob on how to access the properties in C#

890bdc5

adding space between ResetParameters

68fe54f

ervteng reviewed Dec 2, 2019

View reviewed changes

ervteng approved these changes Dec 2, 2019

View reviewed changes

fix typo

214af82

vincentpierre merged commit 83ff9da into develop-side-channel-2 Dec 2, 2019

delete-merged-branch Bot deleted the develop-side-channel-docs branch December 2, 2019 23:09

github-actions Bot locked as resolved and limited conversation to collaborators May 17, 2021

Conversation

vincentpierre commented Dec 2, 2019

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Steps to Migrate

Uh oh!

chriselion left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ervteng Dec 2, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ervteng Dec 2, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ervteng left a comment

Choose a reason for hiding this comment

Uh oh!

MarkTension commented Dec 9, 2019

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

ervteng Dec 2, 2019 •

edited

Loading

ervteng Dec 2, 2019 •

edited

Loading