Roadmap for OpenRL

[huangshiyu13]

This issue will keep tracking of OpenRL's updates of the subsequent few versions:

v1.0.0

• support multi-machine/node training

v0.2.1 (in progress)

• Support Crafter #296
• optimize agent save #15

v0.2.0

• deepspeed support #19 deepspeed support #282
• fix eval_callback bug #268
• fix petting zoo bugs #272
• - add net, gail test #276
• Issue #216: Add envpool to openrl #278

0.1.9

• fix arena petting zoo import error #258
• fix typo: change loss rate to lose rate #256
• add RandomAgent for Arena #262

0.1.8

• fix winning rate output bug of smac #244
• fix atari training bugs #249
• fix windows bugs #250
• add test atari #253

0.1.7

• add smacv2 #235
• fix arena bugs #240

0.1.6

• support loading stable-baseline3's models from hugging face #230

0.1.5

• fix openrl random opponent bugs #224
• add gym_pybullet_drones env #225
• add A2C algorithm #226
• merge dm_control to gymnasium #227

0.1.4

• add JiDi evaluation support #217
• add google footbal evaluation support #218
• add google footbal evaluation support #218

0.1.3

• add custom env examples #212

0.1.2

• add dm-comtrol #201
• add snake environment #209

0.1.1

• add shuffle traj to offline env #187
• support log evaluation callback info to wandb #189
• add arena #190
• add set seed to arena, and test reproducibility #193
• fix y_relu twice written #194

0.1.0

• add selfplay with self #177
• add selfplay strategy #182

0.0.15

• add sac #169
• fix offpolicy bugs #174

0.0.14

• can train tictactoe #158
• add isaac example #161

0.0.13

• add GAIL #130
• add more callbacks and callbacks tests #122
• add behavior cloning algorithm #144

0.0.12

• optimize vdn algorithm v3 #108
• add Atari #106
• Successfully train dqn in gridworld #103
• add StarCraft II: SMAC #109
• can save checkpoints with callbacks #114
• Add connect3 #115
• can evaluate and save best model with callbacks #117

0.0.11

• add MAT algorithm #83
• add mujoco #88

0.0.10

• Add Retro Environment #67 Add retro #73
• update runners for off policy algorithm #71 update module, buffer and driver for off-policy algorithm #74
• Add SuperMario Environment #77
• fix bug in test_case: https://openrl-docs.readthedocs.io/en/latest/quick_start/multi_agent_RL.html#loading-trained-agent #78

0.0.9

• add .vscode to gitignore #55
• - add colab #60
• fix: PPONet parse_args for colab #59
• fix: python version check bugs #58
• abstract rl agent #57

0.0.8

• improve code testing coverage
• - add test for train examples #40
• [Feature Request] add docker image #41
• fix minor bugs

0.0.7

• fix "rewards not found" bug #27
• fix: rewards class info bug for multi agents scenario #29

v0.0.6

• import hugging face models/datasets
• dictionary observation support
• multi-agent training support
• dialog training support
• fix existing bugs
• add more examples
• optimize Contributing docs
• fix existing errors in openrl-docs