DOTS compatible version of ML-Agents
Перейти к файлу
vincentpierre 75d2e49382 Adding a contributing.md file 2021-10-04 13:09:24 -07:00
Documentation~ Updating README 2021-09-22 10:06:48 -07:00
Editor removing old comments 2021-09-22 10:24:50 -07:00
Runtime making pushblock more consistent 2021-09-22 22:07:20 -07:00
Samples~ making pushblock more consistent 2021-09-22 22:07:20 -07:00
Tests Making the master branch compatible with trainers 2.0. Still some features missing 2021-06-07 14:18:44 -07:00
ml-agents-dots-envs~ removing old comments 2021-09-22 10:24:50 -07:00
ml-agents-dots-learn~ Making the master branch compatible with trainers 2.0. Still some features missing 2021-06-07 14:18:44 -07:00
.editorconfig Moved the repo around 2020-02-24 15:39:19 -08:00
.gitignore Fixed the time scale speedup with non-obsolete methods 2021-06-23 12:32:27 -07:00
CHANGELOG.md Update CHANGELOG.md 2020-06-09 15:35:49 -07:00
CHANGELOG.md.meta Added missing meta files 2020-02-24 17:11:48 -08:00
CONTRIBUTING.md Adding a contributing.md file 2021-10-04 13:09:24 -07:00
Editor.meta Moved the repo around 2020-02-24 15:39:19 -08:00
LICENSE.md Modified some formatting and added comments 2020-04-15 14:38:11 -07:00
LICENSE.md.meta Added missing meta files 2020-02-24 17:11:48 -08:00
README.md Updating README 2021-09-22 10:06:48 -07:00
README.md.meta Moved the repo around 2020-02-24 15:39:19 -08:00
Runtime.meta Moved the repo around 2020-02-24 15:39:19 -08:00
Tests.meta Moved the repo around 2020-02-24 15:39:19 -08:00
package.json Updating to 2020.2 and adding a pushblock example 2021-06-22 10:53:28 -07:00
package.json.meta Moved the repo around 2020-02-24 15:39:19 -08:00

README.md

ML-Agents DOTS

This repository is a prototype of a DOTS compatible version of the ML-Agents. It uses the Data Oriented Tech Stack (DOTS) of Unity. Note that not all features of ML-Agents are supported. For example, it is not possible to use imitation learning or cooperative/adversarial behaviors yet. The API of the DOTS version is different from the ML-Agents API in the Editor, it is the responsibility of the user to update the observations of the Agents since there are no implementation of sensors. For more information about the API please see the Documentation.

Documentation