Update README.md
Browse files
README.md
CHANGED
|
@@ -3,6 +3,15 @@ pipeline_tag: reinforcement-learning
|
|
| 3 |
tags:
|
| 4 |
- ppo
|
| 5 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6 |
# Environment
|
| 7 |
Multiplayer pong_v3 from PettingZoo with :
|
| 8 |
- 4 stacked frame
|
|
@@ -46,14 +55,6 @@ def get_env(args, run_name):
|
|
| 46 |
return envs
|
| 47 |
```
|
| 48 |
|
| 49 |
-
# Experiment
|
| 50 |
-
PPO agents trained in a selfplay settings. This repo includes checkpoints collected during training for
|
| 51 |
-
4 experiments:
|
| 52 |
-
- Shared weights for actor and critic
|
| 53 |
-
- No shared weights
|
| 54 |
-
- Resume training for extra steps for both shared and no shared setup
|
| 55 |
-
Please check our [wandb report](https://wandb.ai/dumas/SPAR_RL_ELK/) for more details and the training code on [our GitHub](https://github.com/Butanium/cleanrl/blob/master/multiplayer_pong/ppo_pettingzoo_ma_atari.py)
|
| 56 |
-
|
| 57 |
# Model architecture
|
| 58 |
```py
|
| 59 |
def atari_network(orth_init=False):
|
|
|
|
| 3 |
tags:
|
| 4 |
- ppo
|
| 5 |
---
|
| 6 |
+
# Experiment
|
| 7 |
+
PPO agents trained in a selfplay settings. This repo includes checkpoints collected during training for
|
| 8 |
+
4 experiments:
|
| 9 |
+
- Shared weights for actor and critic
|
| 10 |
+
- No shared weights
|
| 11 |
+
- Resume training for extra steps for both shared and no shared setup
|
| 12 |
+
Please check our [wandb report](https://wandb.ai/dumas/SPAR_RL_ELK/) for more details and the training code on [our GitHub](https://github.com/Butanium/cleanrl/blob/master/multiplayer_pong/ppo_pettingzoo_ma_atari.py)
|
| 13 |
+
|
| 14 |
+
|
| 15 |
# Environment
|
| 16 |
Multiplayer pong_v3 from PettingZoo with :
|
| 17 |
- 4 stacked frame
|
|
|
|
| 55 |
return envs
|
| 56 |
```
|
| 57 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 58 |
# Model architecture
|
| 59 |
```py
|
| 60 |
def atari_network(orth_init=False):
|