## PYSC2 Env
DI-engine uses standard pysc2 env, you can install it as follow:
```shell
pip install pysc2
```
## SMAC Benchmark
==setting: SC2 version=4.6.2.69232, difficulty=7, 2M env step==
| 3s5z | pymarl | |DI-engine | | cfg |
| :----: | :------: | :--: | :------: | :------: | :----------------------------------------------------------: |
| | win rate | time | win rate | time | |
| qmix | 1 | 9.5h | **1** | **3.2h** | dizoo/smac/config/smac_3s5z_qmix_config.py |
| collaq | 1 | 28h | 0.9 | **8.5h** | dizoo/smac/config/smac_3s5z_collaq_config.py |
| coma | 0 | 2.7h | **0.9** | **2.9h** | dizoo/smac/config/smac_3s5z_coma_config.py |
| qtran | 0.1 | 11.5h | **0.9** | **4h** | dizoo/smac/config/smac_3s5z_qtran_config.py |
| ippo | 0.15 | 10.5h | **0.8** | **6.8h** | |
| mappo(ours) | - | - | **1** | **2.4h** | dizoo/smac/config/smac_3s5z_mappo_config.py |
| masac(ours) | - | - | **1** | **4.4h** | dizoo/smac/config/smac_3s5z_masac_config.py |
| 5m_vs_6m | pymarl | |DI-engine | | cfg |
| :-------: | :------: | :--: | :------: | :------: | :----------------------------------------------------------: |
| | win rate | time | win rate | time | |
| qmix | **0.76** | 7.5h | 0.6 | **6.5h** | dizoo/smac/config/smac_5m6m_qmix_config.py |
| collaq | 0.8 | 24h | 0.7 | **9.5h** | dizoo/smac/config/smac_5m6m_collaq_config.py |
| coma | 0 | 2.5h | 0 | - | |
| qtran | 0.7 | 7h | 0.55 | **5.5h** | dizoo/smac/config/smac_5m6m_qtran_config.py |
| ippo | 0 | 9.2h | **0.75** | **6.9h** | |
| mappo(ours) | - | - | **0.75** | **3.2h** | dizoo/smac/config/smac_5m6m_mappo_config.py |
| masac(ours) | - | - | **1** | **5.2h** | dizoo/smac/config/smac_5m6m_masac_config.py |
| MMM | pymarl | |DI-engine | | cfg |
| :----: | :------: | :--: | :------: | :------: | :----------------------------------------------------------: |
| | win rate | time | win rate | time | |
| qmix | 1 | 9.5h | **1** | **3.5h** | dizoo/smac/config/smac_MMM_qmix_config.py |
| collaq | 1 | 38h | **1** | **6.7h** | dizoo/smac/config/smac_MMM_collaq_config.py |
| coma | 0.1 | 3h | **0.9** | **2.6h** | dizoo/smac/config/smac_MMM_coma_config.py |
| qtran | 1 | 8.5h | **1** | **5.5h** | dizoo/smac/config/smac_MMM_qtran_config.py |
| ippo | 0.33 | 7.2h | **1** | **4.7h** | |
| mappo(ours) | - | - | **1** | **2.7h** | dizoo/smac/config/smac_MMM_mappo_config.py |
| masac(ours) | - | - | **1** | **5.2h** | dizoo/smac/config/smac_MMM_masac_config.py |
| MMM2 | pymarl | |DI-engine | | cfg |
| :----: | :------: | :--: | :------: | :------: | :----------------------------------------------------------: |
| | win rate | time | win rate | time | |
| qmix | 0.7 | 10h | 0.4 | **5.5h** | dizoo/smac/config/smac_MMM2_qmix_config.py |
| collaq | 0.9 | 24h | 0.6 | **13h** | dizoo/smac/config/smac_MMM2_collaq_config.py |
| coma | 0 | 3h | **0.2** | 3.5h | dizoo/smac/config/smac_MMM2_coma_config.py |
| qtran | 0 | 8.5h | 0 | - | |
| ippo | 0 | 8.3h | **0.875** | **6h** | |
| mappo(ours) | - | - | **1** | **3.8h** | dizoo/smac/config/smac_MMM2_mappo_config.py |
| masac(ours) | - | - | **1** | **7.2h** | dizoo/smac/config/smac_MMM2_masac_config.py |
| 3s5z_vs_3s6z | MAPPO(Wu) | |DI-engine | | cfg |
| :----: | :------: | :--: | :------: | :------: | :----------------------------------------------------------: |
| | win rate | time | win rate | time | |
| mappo(ours) | - | - | **0.88** | **3.8h** | dizoo/smac/config/smac_3s5zvs3s6z_mappo_config.py |
| masac(ours) | - | - | **1** | **7.2h** | dizoo/smac/config/smac_3s5zvs3s6z_masac_config.py |
| 8m_vs_9m | MAPPO(Wu) | |DI-engine | | cfg |
| :----: | :------: | :--: | :------: | :------: | :----------------------------------------------------------: |
| | win rate | time | win rate | time | |
| mappo(ours) | - | - | **1** | **3.6h** | dizoo/smac/config/smac_3s5zvs3s6z_mappo_config.py |
| masac(ours) | - | - | **1** | **6.7h** | dizoo/smac/config/smac_3s5zvs3s6z_masac_config.py |
| 10m_vs_11m | MAPPO(Wu) | |DI-engine | | cfg |
| :----: | :------: | :--: | :------: | :------: | :----------------------------------------------------------: |
| | win rate | time | win rate | time | |
| mappo(ours) | - | - | **1** | **3.9h** | dizoo/smac/config/smac_10m11m_mappo_config.py |
| masac(ours) | - | - | **1** | **6.9h** | dizoo/smac/config/smac_10m11m_masac_config.py |
| 25m | MAPPO(Wu) | |DI-engine | | cfg |
| :----: | :------: | :--: | :------: | :------: | :----------------------------------------------------------: |
| | win rate | time | win rate | time | |
| mappo(ours) | - | - | **1** | **3.7h** | dizoo/smac/config/smac_25m_mappo_config.py |
| masac(ours) | - | - | **1** | **6.4h** | dizoo/smac/config/smac_25m_masac_config.py |
| 2c_vs_64zg | MAPPO(Wu) | |DI-engine | | cfg |
| :----: | :------: | :--: | :------: | :------: | :----------------------------------------------------------: |
| | win rate | time | win rate | time | |
| mappo(ours) | - | - | **1** | **3.2h** | dizoo/smac/config/smac_2c64zg_mappo_config.py |
| masac(ours) | - | - | **1** | **6.1h** | dizoo/smac/config/smac_2c64zg_masac_config.py |
| corridor | MAPPO(Wu) | |DI-engine | | cfg |
| :----: | :------: | :--: | :------: | :------: | :----------------------------------------------------------: |
| | win rate | time | win rate | time |
DI-engine 是一种广义的决策智能引擎，旨在标准化不同的决策智能环境和应用 还支持各种训练管道和定制的决策 AI 应用程序 （1558个子文件） 
Dockerfile.base 2KB CHANGELOG 16KB .coveragerc 836B Dockerfile.env 4KB .flake8 112B
coinrun.gif 8.48MB overcooked.gif 6.24MB competitive_rl.gif 3.61MB slime_volley.gif 3.32MB maze.gif 3.25MB mujoco.gif 2.27MB minigrid.gif 2.13MB gobigger_overview.gif 2.08MB mario.gif 1.79MB half_offensive.gif 1.37MB moving_v0.gif 1.25MB car_racing.gif 1.16MB atari.gif 1.15MB d4rl.gif 1.14MB gfootball.gif 1.14MB pybullet.gif 1.06MB smac.gif 857KB evogym.gif 649KB gym_pybullet_drones.gif 608KB pendulum.gif 569KB original.gif 397KB petting_zoo_mpe_simple_spread.gif 295KB lunarlander.gif 270KB cartpole.gif 137KB acrobot.gif 102KB bitflip.gif 99KB .gitignore 21KB .gitignore 11B Dockerfile.hpc 2KB pytest.ini 194B
bg.jpg 512KB simple_evironment.json 4KB carry_bot.json 3KB speed_bot.json 2KB LICENSE 11KB Makefile 2KB README.md 8KB README.md 5KB CODE_OF_CONDUCT.md 5KB README.md 5KB README.md 4KB README.md 2KB custom.md 897B README.md 855B README.md 632B README.md 628B README.md 546B README.md 435B README.md 388B README.md 361B README.md 344B readme.md 274B CONTRIBUTING.md 249B PULL_REQUEST_TEMPLATE.md 190B README.md 171B README.md 152B README.md 121B checkpoint-alphapong.pkl 11.47MB checkpoint-strong.pkl 11.47MB checkpoint-medium.pkl 103KB checkpoint-weak.pkl 103KB
beergame.png 1MB maze.png 266KB imagenet.png 218KB parallel_main-sequence.png 200KB serial_main-sequence.png 179KB dmc2gym_cheetah.png 173KB dataloader-sequence.png 170KB serial_collector-activity.png 133KB serial_evaluator-activity.png 81KB bipedalwalkertb.png 74KB coinrun.png 74KB bsuite.png 71KB serial_learner-activity.png 59KB statemachine.png 54KB position.png 51KB wechat.png 38KB league_demo.png 34KB env_state.png 21KB profit.png 14KB target.png 4KB parallel_main-sequence.puml 2KB serial_main.puml 2KB dataloader-sequence.puml 2KB serial_collector-activity.puml 1KB serial_evaluator-activity.puml 702B serial_learner-activity.puml 448B sac.py 80KB smac_env.py 69KB td.py 56KB ppo.py 55KB ppg.py 55KB head.py 54KB football_ikki.py 51KB q_learning.py 50KB共 1558 条
- 1
- 2
- 3
- 4
- 5
- 6
- 16
资源评论
