### Initializing GraphMPEEnv with Wrapper (Python)

Source: https://github.com/nsidn98/informarl/blob/main/README.md

This snippet demonstrates how to initialize the `GraphMPEEnv` using an environment wrapper, which simplifies the setup process by consolidating arguments. It shows the initial reset call, which returns observations, agent IDs, node observations, and adjacency matrices, similar to the direct environment setup.

```python
from multiagent.MPE_env import GraphMPEEnv
# all_args can be pulled config.py or refer to `onpolicy/scripts/train_mpe.py`
env = MPEEnv(all_args)
obs_n, agent_id_n, node_obs_n, adj_n = env.reset()
```

--------------------------------

### Installing Python Dependencies

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/nav/README.md

Installs all required Python packages listed in the `requirements.txt` file using pip, typically executed within a virtual environment to manage dependencies.

```Shell
pip install -r requirements.txt
```

--------------------------------

### Installing Gym Flock Environment (Shell)

Source: https://github.com/nsidn98/informarl/blob/main/baselines/gpg/gym_formation/README.md

This shell command installs the Gym Flock environment in editable mode, making it accessible for development and use within Python projects. It requires Python 3 and pip3 to be installed on the system.

```Shell
pip3 install -e .
```

--------------------------------

### Installing Gym Vecenv from Git Repository

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/requirements.txt

This entry installs the 'gym_vecenv' package directly from a Git repository, specifically from the 'smac' branch, allowing for editable installation of a custom version.

```Python
-e git+https://github.com/agakshat/gym_vecenv.git@smac#egg=gym_vecenv
```

--------------------------------

### Installing nomkl to Resolve OMP Error (Shell)

Source: https://github.com/nsidn98/informarl/blob/main/README.md

Resolves `OMP: Error #15` by installing the `nomkl` package, which prevents conflicts when `libiomp5.dylib` is initialized while `libomp.dylib` is already present. This is a common fix for MKL-related library conflicts.

```Shell
conda install nomkl
```

--------------------------------

### Installing Multiagent from Git Repository

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/requirements.txt

This entry installs the 'multiagent' package from a specific commit (123a7431aa5eb085194d9d1b586c7eb01764c651) within the 'marl_transfer' Git repository, targeting the 'mape' subdirectory for an editable installation.

```Python
-e git+https://github.com/sumitsk/marl_transfer@123a7431aa5eb085194d9d1b586c7eb01764c651#egg=multiagent&subdirectory=mape
```

--------------------------------

### Installing TensorboardX Python Dependency

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/requirements.txt

This entry defines the 'tensorboardX' package and fixes its version to 1.7, providing TensorBoard visualizations for PyTorch.

```Python
tensorboardX==1.7
```

--------------------------------

### Installing Gym Python Dependency

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/requirements.txt

This line specifies the 'gym' package with version 0.10.8, a toolkit for developing and comparing reinforcement learning algorithms.

```Python
gym==0.10.8
```

--------------------------------

### Training GPG Model from Scratch - Python

Source: https://github.com/nsidn98/informarl/blob/main/baselines/gpg/rl_navigation/README.md

This command initiates the training process for the GPG model from scratch. It executes the `main.py` script, which encapsulates the core training logic. Successful execution requires prior installation of the Deep Graph Library (DGL) and a compatible version of PyTorch.

```Shell
python main.py
```

--------------------------------

### Training InforMARL with RMAPPO (Bash)

Source: https://github.com/nsidn98/informarl/blob/main/README.md

This bash command initiates the training process for InforMARL using the RMAPPO algorithm. It configures various parameters such as project name, environment, number of agents, episode length, and learning rates, enabling the model to learn navigation policies in a graph-based multi-agent environment.

```bash
python -u onpolicy/scripts/train_mpe.py --use_valuenorm --use_popart \
--project_name "informarl" \
--env_name "GraphMPE" \
--algorithm_name "rmappo" \
--seed 0 \
--experiment_name "informarl" \
--scenario_name "navigation_graph" \
--num_agents 3 \
--collision_rew 5 \
--n_training_threads 1 --n_rollout_threads 128 \
--num_mini_batch 1 \
--episode_length 25 \
--num_env_steps 2000000 \
--ppo_epoch 10 --use_ReLU --gain 0.01 --lr 7e-4 --critic_lr 7e-4 \
--user_name "marl" \
--use_cent_obs "False" \
--graph_feat_type "relative" \
--auto_mini_batch_size --target_mini_batch_size 128
```

--------------------------------

### Installing Chardet Python Dependency

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/requirements.txt

This entry defines the 'chardet' package and fixes its version to 3.0.4, used for character encoding detection.

```Python
chardet==3.0.4
```

--------------------------------

### Installing Pyglet Python Dependency

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/requirements.txt

This line specifies the 'pyglet' package with version 1.3.2, a cross-platform windowing and multimedia library for Python.

```Python
pyglet==1.3.2
```

--------------------------------

### Installing Python Package from Git Repository (Editable)

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/nav/requirements.txt

This snippet illustrates how to install a Python package directly from a Git repository in editable mode. This is particularly useful during development, allowing immediate reflection of changes made to the package's source code without requiring reinstallation. It specifies a branch or tag for the installation.

```Python
-e git+https://github.com/agakshat/gym_vecenv.git@smac#egg=gym_vecenv
```

--------------------------------

### Setting Up and Running Graph Navigation Environment (Python)

Source: https://github.com/nsidn98/informarl/blob/main/README.md

This Python code demonstrates how to initialize and interact with the `MultiAgentGraphEnv`, a custom environment designed for Graph Neural Networks. It defines environment parameters, creates the world and environment instances, sets up interactive policies for agents, and runs an execution loop to step through the environment, rendering agent views.

```python
from multiagent.environment import MultiAgentGraphEnv
from multiagent.policy import InteractivePolicy

# makeshift argparser
class Args:
    def __init__(self):
        self.num_agents:int=3
        self.world_size=2
        self.num_scripted_agents=0
        self.num_obstacles:int=3
        self.collaborative:bool=False 
        self.max_speed:Optional[float]=2
        self.collision_rew:float=5
        self.goal_rew:float=5
        self.min_dist_thresh:float=0.1
        self.use_dones:bool=False
        self.episode_length:int=25
        self.max_edge_dist:float=1
        self.graph_feat_type:str='global'
args = Args()

scenario = Scenario()
# create world
world = scenario.make_world(args)
# create multiagent environment
env = MultiAgentGraphEnv(world=world, reset_callback=scenario.reset_world, 
                    reward_callback=scenario.reward, 
                    observation_callback=scenario.observation, 
                    graph_observation_callback=scenario.graph_observation,
                    info_callback=scenario.info_callback, 
                    done_callback=scenario.done,
                    id_callback=scenario.get_id,
                    update_graph=scenario.update_graph,
                    shared_viewer=False)
# render call to create viewer window
env.render()
# create interactive policies for each agent
policies = [InteractivePolicy(env,i) for i in range(env.n)]
# execution loop
obs_n, agent_id_n, node_obs_n, adj_n = env.reset()
stp=0
while True:
    # query for action from each agent's policy
    act_n = []

    for i, policy in enumerate(policies):
        act_n.append(policy.action(obs_n[i]))
    # step environment
    obs_n, agent_id_n, node_obs_n, adj_n, reward_n, done_n, info_n = env.step(act_n)

    # render all agent views
    env.render()
```

--------------------------------

### Initiating Curriculum Training with Python

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/README.md

This command starts curriculum training for the 'simple_spread' environment. It uses `automate.py` to manage the curriculum, enables entity-level message passing via `--entity-mp`, and saves results to directory `0`.

```Python
python automate.py --env-name simple_spread --entity-mp --save-dir 0
```

--------------------------------

### Installing Python Package from Git Repository with Subdirectory (Editable)

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/nav/requirements.txt

This snippet shows how to install a Python package from a Git repository in editable mode when the package's root is located within a subdirectory of the repository. It includes a specific commit hash for precise version control and the 'subdirectory' parameter to correctly locate the package.

```Python
-e git+https://github.com/sumitsk/marl_transfer@123a7431aa5eb085194d9d1b586c7eb01764c651#egg=multiagent&subdirectory=mape
```

--------------------------------

### Training Formation Control with Python

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/README.md

This command starts normal training for the 'simple_formation' (formation control) environment with 3 agents. The `--save-dir 0` flag indicates the directory where training outputs will be stored.

```Python
python main.py --env-name simple_formation --num-agents 3  --save-dir 0
```

--------------------------------

### Installing Idna Python Dependency

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/requirements.txt

This line specifies the 'idna' package with version 2.8, used for Internationalized Domain Names in Applications.

```Python
idna==2.8
```

--------------------------------

### Starting Curriculum Training for simple_spread in Python

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/nav/README.md

Executes curriculum training for the 'simple_spread' environment, enabling entity message passing and saving results to directory 0. Before running, the number of agents must be specified within the `automate.py` file.

```Python
python automate.py --env-name simple_spread --entity-mp --save-dir 0
```

--------------------------------

### Installing Cloudpickle Python Dependency

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/requirements.txt

This line specifies the 'cloudpickle' package with version 1.2.1, which is used for serializing Python objects.

```Python
cloudpickle==1.2.1
```

--------------------------------

### Installing Six Python Dependency

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/requirements.txt

This line specifies the 'six' package with version 1.12.0, a Python 2 and 3 compatibility library.

```Python
six==1.12.0
```

--------------------------------

### Installing Urllib3 Python Dependency

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/requirements.txt

This entry defines the 'urllib3' package and fixes its version to 1.26.5, a powerful, user-friendly HTTP client for Python.

```Python
urllib3==1.26.5
```

--------------------------------

### Installing Requests Python Dependency

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/requirements.txt

This line specifies the 'requests' package with version 2.22.0, an elegant and simple HTTP library for Python.

```Python
requests==2.22.0
```

--------------------------------

### Installing Pkg-Resources Python Dependency

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/requirements.txt

This line specifies the 'pkg-resources' package with version 0.0.0, typically part of setuptools, used for package resource management.

```Python
pkg-resources==0.0.0
```

--------------------------------

### Installing Protobuf Python Dependency

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/requirements.txt

This entry defines the 'protobuf' package and fixes its version to 3.8.0, used for Google's data interchange format.

```Python
protobuf==3.8.0
```

--------------------------------

### Installing SciPy Python Dependency

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/requirements.txt

This entry defines the 'scipy' package and fixes its version to 1.3.0, a fundamental library for scientific and technical computing.

```Python
scipy==1.3.0
```

--------------------------------

### Installing Python-Utils Python Dependency

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/requirements.txt

This entry defines the 'python-utils' package and fixes its version to 2.3.0, providing various utility functions for Python.

```Python
python-utils==2.3.0
```

--------------------------------

### Executing Graph Navigation Scenario (Python)

Source: https://github.com/nsidn98/informarl/blob/main/README.md

This command directly executes the `navigation_graph.py` script, which sets up and runs the Graph Neural Network compatible navigation environment. It's a standalone way to observe the environment's behavior without full training.

```python
python multiagent/custom_scenarios/navigation_graph.py
```

--------------------------------

### Installing NumPy Python Dependency

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/requirements.txt

This line specifies the 'numpy' package and pins its version to 1.22.0, a fundamental package for scientific computing with Python.

```Python
numpy==1.22.0
```

--------------------------------

### Installing Torch Python Dependency

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/requirements.txt

This line specifies the 'torch' package with version 1.1.0, the core library for PyTorch, a deep learning framework.

```Python
torch==1.1.0
```

--------------------------------

### Installing Certifi Python Dependency

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/requirements.txt

This line specifies the 'certifi' package and pins its version to 2019.6.16, ensuring consistent dependency resolution for SSL certificate validation.

```Python
certifi==2019.6.16
```

--------------------------------

### Training Formation Control (simple_formation) in Python

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/nav/README.md

Starts normal training for the 'simple_formation' environment with 3 agents. This command saves the training progress and results to the specified directory 0, without enabling entity message passing.

```Python
python main.py --env-name simple_formation --num-agents 3  --save-dir 0
```

--------------------------------

### Installing NumPy-STL Python Dependency

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/requirements.txt

This entry defines the 'numpy-stl' package and fixes its version to 2.10.1, used for reading and writing STL files with NumPy.

```Python
numpy-stl==2.10.1
```

--------------------------------

### Installing Future Python Dependency

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/requirements.txt

This entry defines the 'future' package and fixes its version to 0.17.1, providing Python 3 compatibility to Python 2 code.

```Python
future==0.17.1
```

--------------------------------

### Reloading GPG Parameters for Formation Viewing - Python

Source: https://github.com/nsidn98/informarl/blob/main/baselines/gpg/rl_navigation/README.md

This command reloads pre-trained GPG model parameters, primarily for visualizing formations within the navigation environment. The `parameter_reload.py` script is responsible for loading model data, which it retrieves from the `./logs` directory. This operation also depends on DGL and PyTorch being installed.

```Shell
python parameter_reload.py
```

--------------------------------

### Listing Python Project Dependencies

Source: https://github.com/nsidn98/informarl/blob/main/requirement.txt

This snippet specifies the exact versions of Python packages required for the project. These dependencies are typically installed using pip from a `requirements.txt` file to ensure environment consistency and reproducibility across different development and deployment environments.

```Python
gym==0.10.5
numpy==1.19.4
pyglet==1.5.26
protobuff==3.20.0
PyOpenGL==3.1.5
PyQt5==5.9.2
requests==2.22.0
tensorboard==2.4.0
tensorboard-plugin-wit==1.7.0
tensorboardX==2.1
torch==1.8.1
torch-cluster==1.5.9
torch-geometric==2.0.4
torch-scatter==2.0.7
torch-sparse==0.6.10
torch-spline-conv==1.2.1
torchvision==0.9.1
tqdm==4.44.1
typing==3.6.4
typing-extensions==3.7.4.3
wandb==0.10.31
```

--------------------------------

### Citing the informarl codebase (BibTeX)

Source: https://github.com/nsidn98/informarl/blob/main/README.md

Provides the BibTeX entry for citing the `informarl` codebase, which refers to the article 'Scalable Multi-Agent Reinforcement Learning through Intelligent Information Aggregation' by Nayak et al. This citation is useful for researchers who utilize the codebase in their work.

```BibTeX
@article{nayak22informarl,
  doi = {10.48550/ARXIV.2211.02127},
  url = {https://arxiv.org/abs/2211.02127},
  author = {Nayak, Siddharth and Choi, Kenneth and Ding, Wenqi and Dolan, Sydney and Gopalakrishnan, Karthik and Balakrishnan, Hamsa},
  keywords = {Multiagent Systems (cs.MA), Artificial Intelligence (cs.AI), Robotics (cs.RO), FOS: Computer and information sciences, FOS: Computer and information sciences},
  title = {Scalable Multi-Agent Reinforcement Learning through Intelligent Information Aggregation},
  publisher = {arXiv},
  year = {2022},
  copyright = {Creative Commons Attribution 4.0 International}
}
```

--------------------------------

### Upgrading pyglet for MacOS Big Sur (Shell)

Source: https://github.com/nsidn98/informarl/blob/main/README.md

Provides a hacky fix for `AttributeError: dlsym(RTLD_DEFAULT, CFStringCreateWithCString): symbol not found` on MacOS Big Sur. This is achieved by upgrading `pyglet` to its `1.5-maintenance` branch directly from GitHub, addressing a symbol not found error.

```Shell
pip install --user --upgrade git+http://github.com/pyglet/pyglet@pyglet-1.5-maintenance
```

--------------------------------

### Reinstalling torch-geometric with specific CUDA version (Shell)

Source: https://github.com/nsidn98/informarl/blob/main/README.md

Addresses `AttributeError: 'NoneType' object has no attribute 'origin'` when using `torch-geometric` with CUDA. This involves uninstalling related `torch` packages and then reinstalling `torch-scatter`, `torch-sparse`, and `torch-geometric` from specific URLs, ensuring compatibility with a defined `TORCH` and `CUDA` version.

```Shell
TORCH="1.8.0"
CUDA="cu102"
pip install --no-index torch-scatter -f https://data.pyg.org/whl/torch-${TORCH}+${CUDA}.html --user
pip install --no-index torch-sparse -f https://data.pyg.org/whl/torch-${TORCH}+${CUDA}.html --user
pip install torch-geometric --user
```

--------------------------------

### Initializing Gym Flock FormationFlying-v3 Environment (Python)

Source: https://github.com/nsidn98/informarl/blob/main/baselines/gpg/gym_formation/README.md

This Python snippet demonstrates how to import the necessary Gym and gym_flock libraries and instantiate the 'FormationFlying-v3' environment. Once initialized, the environment can be controlled using standard OpenAI Gym methods like env.reset() and env.step(), and also provides env.controller() for imitation learning.

```Python
import gym
import gym_flock
env = gym.make("FormationFlying-v3")
```

--------------------------------

### Running MPNN Navigation Baseline (Shell)

Source: https://github.com/nsidn98/informarl/blob/main/baselines/README.md

This command initiates the MPNN navigation baseline, configuring it with 128 rollout threads, a 'navigation' scenario, and global observations. It also enables verbose output and integration with Weights & Biases for experiment tracking.

```Shell
python baselines/mpnn/nav/main.py --n_rollout_threads=128 --scenario_name='navigation' --use_wandb --verbose --obs_type 'global'
```

--------------------------------

### Executing GPG Reinforcement Learning (Shell)

Source: https://github.com/nsidn98/informarl/blob/main/baselines/README.md

This command executes the GPG reinforcement learning baseline for the MPE environment. It configures the experiment with specific parameters such as project name, environment, algorithm, seed, graph type, scenario, number of agents, and total environment steps, enabling tracking with Weights & Biases.

```Shell
python -u -W ignore baselines/gpg/rl_navigation/main.py \
--project_name "compare_3" \
--env_name "MPE" \
--algorithm_name "gpg" \
--seed 0 \
--graph_type "static" \
--experiment_name "test" \
--scenario_name "navigation_gpg" \
--num_agents=3 \
--num_env_steps 2000000 \
--user_name "marl" --use_wandb
```

--------------------------------

### Running CADRL Navigation Baseline (Shell)

Source: https://github.com/nsidn98/informarl/blob/main/baselines/README.md

This snippet shows the main entry point for running the CADRL navigation baseline. It executes a Python script located at `baselines/cadrl/cadrl_navigation/main.py`, indicating the primary execution file for the CADRL algorithm.

```Shell
python baselines/cadrl/cadrl_navigation/main.py
```

--------------------------------

### Training Line Control (simple_line) in Python

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/nav/README.md

Launches normal training for the 'simple_line' environment with 3 agents. The training output and models are saved to directory 0, following the standard training procedure.

```Python
python main.py --env-name simple_line --num-agents 3  --save-dir 0
```

--------------------------------

### Training Value-Based MARL Algorithms (Shell)

Source: https://github.com/nsidn98/informarl/blob/main/baselines/README.md

This script trains various value-based multi-agent reinforcement learning algorithms (MVDN, VDN, MQMIX, QMIX) within the MPE environment. It sets parameters like the number of agents and landmarks, episode length, learning rate, and enables soft updates and reward normalization, with results tracked via Weights & Biases.

```Shell
algo="mvdn" # or "vdn", "mqmix", "qmix"

python baselines/offpolicy/scripts/train/train_mpe.py \
--env_name "MPE" \
--algorithm_name ${algo} --n_rollout_threads=1 \
--experiment_name "test" \
--scenario_name "navigation" \
--num_agents 3 \
--num_landmarks 3 \
--seed 0 \
--episode_length 25 \
--use_soft_update \
--lr 7e-4 --use_reward_normalization \
--hard_update_interval_episode 200 \
--num_env_steps 30000 --use_wandb \
--obs_type='nbd'
```

--------------------------------

### Training Line Control with Python

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/README.md

This command launches normal training for the 'simple_line' (line control) environment, configuring it for 3 agents. The `--save-dir 0` argument specifies the output directory for saving training data.

```Python
python main.py --env-name simple_line --num-agents 3  --save-dir 0
```

--------------------------------

### Training Coverage Control with Python

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/README.md

This command initiates normal training for the 'simple_spread' (coverage control) environment using 3 agents. The `--entity-mp` flag enables entity-level message passing, and `--save-dir 0` specifies the directory for saving training results.

```Python
python main.py --env-name simple_spread --num-agents 3 --entity-mp --save-dir 0
```

--------------------------------

### BibTeX Citation for Emergence of Grounded Compositional Language (arXiv 2017)

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/mape/README.md

BibTeX entry for the 'Emergence of Grounded Compositional Language in Multi-Agent Populations' paper, which describes the original particle world environment. This citation is useful for academic referencing.

```BibTeX
@article{mordatch2017emergence,
  title={Emergence of Grounded Compositional Language in Multi-Agent Populations},
  author={Mordatch, Igor and Abbeel, Pieter},
  journal={arXiv preprint arXiv:1703.04908},
  year={2017}
}
```

--------------------------------

### Transferring Policy for Multi-Agent Training with Python

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/README.md

This command demonstrates how to continue training a policy by loading a pre-trained model. It loads a model trained with 3 agents (`models/ss/na3_uc.pt`) and continues training for a team of 5 agents in the 'simple_spread' environment, enabling transfer learning.

```Python
python main.py --env-name simple_spread --entity-mp --continue-training --load-dir models/ss/na3_uc.pt --num-agents 5
```

--------------------------------

### Training Actor-Critic MARL Algorithms (Shell)

Source: https://github.com/nsidn98/informarl/blob/main/baselines/README.md

This script trains various actor-critic multi-agent reinforcement learning algorithms (MADDPG, RMADDPG, MATD3, RMATD3) in the MPE environment. It configures training with specific parameters for agents, landmarks, episode length, learning rate, and update intervals, integrating with Weights & Biases for experiment management.

```Shell
algo="maddpg" # or "rmaddpg", "matd3", "rmatd3"

python baselines/offpolicy/scripts/train/train_mpe.py \
--env_name "MPE" \
--algorithm_name ${algo} \
--experiment_name "test" \
--scenario_name "navigation" \
--num_agents 3 \
--num_landmarks 3 \
--seed 0 \
--actor_train_interval_step 1 \
--episode_length 25 \
--use_soft_update \
--lr 7e-4 \
--hard_update_interval_episode 200 \
--num_env_steps 10000000 --use_wandb
```

--------------------------------

### BibTeX Citation for Multi-Agent Actor-Critic (NIPS 2017)

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/mpe/mape/README.md

BibTeX entry for the 'Multi-Agent Actor-Critic for Mixed Cooperative-Competitive Environments' paper, which describes the environments in this repository. This citation is useful for academic referencing.

```BibTeX
@article{lowe2017multi,
  title={Multi-Agent Actor-Critic for Mixed Cooperative-Competitive Environments},
  author={Lowe, Ryan and Wu, Yi and Tamar, Aviv and Harb, Jean and Abbeel, Pieter and Mordatch, Igor},
  journal={Neural Information Processing Systems (NIPS)},
  year={2017}
}
```

--------------------------------

### BibTeX Citation for Graph Convolutional Reinforcement Learning

Source: https://github.com/nsidn98/informarl/blob/main/baselines/dgn/README.md

This BibTeX entry provides the citation details for the 'Graph Convolutional Reinforcement Learning' paper, which describes the DGN model implemented in this project. It includes authors, title, publication venue, and year, suitable for academic referencing.

```BibTeX
	@inproceedings{jiang2020graph,
	    	title={Graph Convolutional Reinforcement Learning},
	    	author={Jiang, Jiechuan and Dun, Chen and Huang, Tiejun and Lu, Zongqing},
	    	booktitle={ICLR},
	    	year={2020}
	}
```

--------------------------------

### Training Coverage Control (simple_spread) in Python

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/nav/README.md

Initiates normal training for the 'simple_spread' environment with 3 agents, enabling entity message passing and saving results to directory 0. The `--save-dir 0` flag indicates where to store the training outputs.

```Python
python main.py --env-name simple_spread --num-agents 3 --entity-mp --save-dir 0
```

--------------------------------

### Continuing Training with Transfer Learning in Python

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/nav/README.md

Continues training for the 'simple_spread' environment with 5 agents, loading a pre-trained policy from `models/ss/na3_uc.pt`. This enables transfer learning from a model trained with 3 agents to a team of 5 agents, with entity message passing enabled.

```Python
python main.py --env-name simple_spread --entity-mp --continue-training --load-dir models/ss/na3_uc.pt --num-agents 5
```

--------------------------------

### Specifying Standard Python Package Version

Source: https://github.com/nsidn98/informarl/blob/main/baselines/mpnn/nav/requirements.txt

This snippet demonstrates how to declare a standard Python package dependency with an exact version. This practice ensures reproducibility across different environments by pinning the package to a specific release, preventing unexpected breaking changes from newer versions.

```Python
certifi==2019.6.16
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.