0.1.3

Highlights

This is the last tensorflow system release. After this, tensorflow systems will be deprecated in favour of Jax systems and our new callback redesign (#457).

Systems

• Updates to acme, reverb and tensorflow.
• Working centralised and state based architectures.
• Recurrent and Multiple Trainer PPO.

Environments

What's Changed

• Bugfix/ Release aren't triggering pypi push job. by @KaleabTessera in #466
• Feature / Release 0.1.2 v2 by @KaleabTessera in #467
• fix: Update black version. by @DriesSmit in #470
• Bugfix/ Update PZ Version and new jax dockerfiles by @KaleabTessera in #480
• Feature/recurrent and multiple trainer MAPPO by @DriesSmit in #326
• Feat/maddpg obs optim by @AsadJeewa in #459
• feat: Add fixed sampler capability + bugfixes by @DriesSmit in #475
• Feature/fix sampler madqn by @EdanToledo in #477
• chore: Up the patch version of mava - 0.1.3. by @KaleabTessera in #485
• Bugfix/fix old tf architectures by @KaleabTessera in #552
• Release 0.1.3 by @KaleabTessera in #486

Full Changelog: 0.1.2...0.1.3

0.1.2

Highlights

Systems

• Fixed observation network bug in mappo + changed implementation to use two optims.
• Fixes in maddpg/mad4pg loss calculation.
• Began on jax system implementations.

Environments

What's Changed

• Fix/add loss mask to ppo by @EdanToledo in #441
• Mainetenance: Fix tf examples issues by @AsadJeewa in #444
• fix: shared weights with agent type by @AsadJeewa in #428
• Fix broken readme links and neaten up formatting by @AsadJeewa in #446
• Feature/jax abstract builder class by @arnupretorius in #433
• docs: updated docs to better represent available options by @sash-a in #448
• Feature/jax general system class by @arnupretorius in #425
• Bugfix/Mypy Inconsistency Issue by @KaleabTessera in #458
• fix/remove flatland wrapper debug print statement by @mmorris44 in #456
• Feature/MAPPO Obs Networks Fix + Multiple Optims by @KaleabTessera in #454
• Feature/new issue template for investigations by @KaleabTessera in #461
• Bugfix/MADD(4)PG by @DriesSmit in #460
• Feat/Upped pypi version. by @KaleabTessera in #464
• Feature / Release 0.1.2 by @KaleabTessera in #465

Full Changelog: 0.1.1...0.1.2

0.1.1

Highlights

Systems

• Stable versions of all systems - noteably stable mappo, vdn and qmix.
• Multiple trainer implementations for maddpg and mad4pg.
• Removed the dial system.

Environments/ Environment Wrappers

• Added Melting Pot support.

What's Changed

• Feature/Enforce docstring code coverage. by @KaleabTessera in #271
• Chore/Resized gifs in readme. by @KaleabTessera in #272
• Feature/Improve Mava agent networks by @DriesSmit in #258
• Feature/upgrade acme version and use new adders by @KaleabTessera in #274
• Chore/Updated makefile and readme for Windows. by @Nashlen in #273
• Fix/supersuit version by @KaleabTessera in #277
• Chore/ Update quickstart by @KaleabTessera in #278
• Feature/New acme adders and tests. by @KaleabTessera in #276
• feature: working version of importance sampling on feedforward madqn. by @jcformanek in #275
• fix/ Smac Load by @KaleabTessera in #283
• update Dockerfile for SMAC installation by @mnguyen0226 in #286
• Bugfix: Simple_spread observation code. by @DriesSmit in #288
• Bugfix/launchpad flag issue by @KaleabTessera in #291
• Feature/mava reproducibility and PZ wrapper fix by @KaleabTessera in #296
• fix: Autorom manual download. by @KaleabTessera in #300
• Feature: Add Readme for setting up a new environment by @DriesSmit in #299
• Chore/re add autorom by @KaleabTessera in #302
• Add checkpoint save interval variable. by @DriesSmit in #301
• Feature/Upgraded tf and reverb versions. by @KaleabTessera in #303
• Chore/flatland gif by @arnupretorius in #304
• Small readme updates by @arnupretorius in #305
• Feature: added rendering to flatland wrapper. by @jcformanek in #307
• chore/Updates for new acme version. by @KaleabTessera in #308
• Fix per agent loggers by @DriesSmit in #313
• Removed deprecated shared_weights parameter by @mmorris44 in #319
• docs: update README with correct link by @AsadJeewa in #320
• small fix for README.md by @arnupretorius in #322
• Feature/Multiple trainers for MA-DDPG by @DriesSmit in #253
• Fix Flatland package error in Docker build by @DriesSmit in #328
• Feature/melting pot by @ldfrancis in #324
• Fix RoboCup environment wrapper by @DriesSmit in #334
• Feature/eval intervals by @KaleabTessera in #323
• Feature/ Smac wrapper Update, MADQN/QMIX/VDN upgrades and Dockerfile improvements by @KaleabTessera in #310
• Feature/add robocup gif by @DriesSmit in #336
• Feature/auto-push-docker-images and version upgrades by @KaleabTessera in #342
• Added a brief explanation of Logging metrics to README by @RuanJohn in #341
• Updated pip installation instructions in README by @RuanJohn in #343
• Bugfix/dockerfile no module found by @KaleabTessera in #344
• feat(git): Added feature and bug templates. by @KaleabTessera in #350
• Doc/meltingpot gif by @ldfrancis in #351
• Replace types ParallelAdder with ReverbParallelAdder by @AsadJeewa in #356
• Update README to link to pypi package by @AsadJeewa in #360
• Feature/auto docs by @KaleabTessera in #354
• Feature/maintainace issue template by @arnupretorius in #368
• Fix/broken launchpad link by @sash-a in #370
• feat: filter docker image push based on label. by @KaleabTessera in #375
• Maintenance/update readme by @arnupretorius in #378
• chore: expand code owner list for better code review by @arnupretorius in #390
• Feature/internal feature issue template by @arnupretorius in #379
• Feature/internal bug issue template by @arnupretorius in #381
• feat: benchmarking issue template by @arnupretorius in #385
• Bugfix: Fixed conventional commit pre-commit hook not running. by @KaleabTessera in #395
• Fix/checklist for issue templates by @arnupretorius in #388
• feat: internal issue tempalte for tests by @arnupretorius in #399
• chore: add optional benchmark questions to feature by @arnupretorius in #401
• Fix/madqn by @jcformanek in #362
• Fix/architecture typo fix. by @RuanJohn in #410
• Fix/Smac Wrapper Relies on Flatland Installation by @KaleabTessera in #413
• refactor: move examples into tf folder and update examples links by @arnupretorius in #416
• fix: readd quickstart notebook by @arnupretorius in #417
• fix: Fix broken tests due to new gym version. by @KaleabTessera in #421
• Maintenance: Remove redundant value_network code by @AsadJeewa in #423
• fix: small bug in the pettingzoo wrapper related to legal action masking by @jcformanek in #432
• Fix/Flatland Docker Container by @KaleabTessera in #437
• Feature/jax abstract system class by @arnupretorius in #405
• Feature/ppo multiple train steps by @EdanToledo in #353
• Fix/ Fix docs build. by @KaleabTessera in #435
• Feature/jax mava custom config class by @arnupretorius in #414
• Feat/Release new mava version. by @KaleabTessera in #438
• Merge: Merge Dev into Main for Release by @KaleabTessera in #439

New Contributors

• @Nashlen made their first contribution in #273
• @mnguyen0226 made their first contribution in #286
• @mmorris44 made their first contribution in #319
• @AsadJeewa made their first contribution in #320
• @RuanJohn made their first contribution in #341
• @sash-a made their first contribution in #370
• @EdanToledo made their first contribution in #353

Full Changelog: 0.1.0...0.1.1

Mava-0.1.0

Highlights

Mava Core

• Components

  • Architectures
    • Added Centralised, Decentralised, Networked and State Based Architectures.
  • Modules
    • Added Broadcast Communication, Epsilon Decay Scheduling, Additive and Monotonic Mixing and Fingerprint Stabilization.
  • Networks
    • Added Additive and Monotic Mixing Networks, Hypernetworks, Communication Networks, Epsilon Greedy and DiscreteValued head.

• Environment Loops

  • Added Parallel and Sequential Environment Loops.

• Adders

  • Added Parallel versions of Transition, Sequential and Episode Adders.

Systems

• Added feedforward training for maddpg, mad4pg, madqn, mappo, vdn and qmix.
• Added recurrent training for madqn, dial, maddpg and mad4pg.
• Added continuous network heads for maddpg, mad4pg and mappo.
• Added decentralised architecture training for maddpg, mad4pg, madqn, mappo, dial, vdn and qmix.
• Added centralised architecture training for maddpg, mad4pg and mappo.
• Added state based architecture training for maddpg and mad4pg.
• Added networked architecture training for maddpg.

Environments/ Environment Wrappers

• Added PettingZoo, SMAC, RoboCup, OpenSpiel, Flatland, Debug Simple Spread, Debug Switch environment and Debug Two-Step game.

Examples

• Added quickstart notebook.
• Added basic examples for sample systems and environments.

Minor Changes and Fixes

MAVA Pre-release

MAVA Pre-release test before official 0.1.0 release.