README.md

---
license: mit
library_name: marl-ppo-suite
tags:
- reinforcement-learning
- starcraft-mac
- smacv2
- mappo
- terran_10_vs_10
- smacv2_terran_10_vs_10
model-index:
- name: MAPPO on smacv2_terran_10_vs_10
  results:
  - task:
      type: reinforcement-learning
      name: StarCraft Multi-Agent Challenge v2
    dataset:
      name: terran_10_vs_10
      type: smacv2
    metrics:
      - name: win-rate
        type: win_rate
        value: 0.566
      - name: mean-reward
        type: mean_reward
        value: 16.68
      - name: mean-ep-length
        type: mean_episode_length
        value: 51.4
---

# MAPPO on **smacv2_terran_10_vs_10**

*10 M environment steps · 14.47 h wall-clock · seed 1*

This is a trained model of a `MAPPO` agent playing *smacv2_terran_10_vs_10*.  
The model was produced with the open-source
[`marl-ppo-suite`](https://github.com/legalaspro/marl-ppo-suite) training
code.



## Usage  – quick evaluation / replay

```bash
# 1. install the codebase (directly from GitHub)
pip install "marl-ppo-suite @ git+https://github.com/legalaspro/marl-ppo-suite"

# 2. get the weights & config from HF
wget https://huggingface.co/<repo-id>/resolve/main/final-torch.model
wget https://huggingface.co/<repo-id>/resolve/main/config.json

# 3-a. Generate a StarCraft II replay file 1 episode in starcraft replay folder
marl-train   --mode render   --model final-torch.model   --config config.json   --render_episodes 1 \       

# 3-b. generate additionally video drawn from frames 
marl-train   --mode render   --model final-torch.model   --config config.json   --render_episodes 1   --render_mode rgb_array       
```

## Files

* **`final-torch.model`** – PyTorch checkpoint  
* **`replay.mp4`** – gameplay of the final policy 
* **`config.json`** – training config
* **`tensorboard/`** – full logs

## Hyper-parameters

```python
{
  "clip_param": 0.05,
  "data_chunk_length": 10,
  "entropy_coef": 0.01,
  "fc_layers": 2,
  "gae_lambda": 0.95,
  "gamma": 0.99,
  "hidden_size": 64,
  "lr": 0.0005,
  "n_steps": 200,
  "num_mini_batch": 1,
  "ppo_epoch": 5,
  "reward_norm_type": "efficient",
  "seed": 1,
  "state_type": "AS",
  "use_reward_norm": true,
  "use_rnn": true,
  "use_value_norm": false,
  "value_norm_type": "welford"
}
```