Getting Started#

Installation#

Install from PyPI:

pip install torchwm

Install from source:

git clone https://github.com/ParamThakkar123/torchwm.git
cd torchwm
pip install -e .

For development and tests:

pip install -e ".[dev]"

Docker#

Build the Docker image:

docker build -t torchwm .

Run the container:

docker run -p 8000:8000 torchwm

Docker Compose#

Build and run all services:

docker compose up -d

This starts both the backend (port 8000) and frontend (port 5173) services.

Quick Start: Dreamer#

from world_models.models import DreamerAgent
from world_models.configs import DreamerConfig

cfg = DreamerConfig()
cfg.env_backend = "gym"
cfg.env = "Pendulum-v1"
cfg.total_steps = 10_000

agent = DreamerAgent(cfg)
agent.train()

Quick Start: JEPA#

from world_models.models import JEPAAgent
from world_models.configs import JEPAConfig

cfg = JEPAConfig()
cfg.dataset = "imagefolder"
cfg.root_path = "./data"
cfg.image_folder = "train"
cfg.epochs = 10

agent = JEPAAgent(cfg)
agent.train()

Environment Backends#

Dreamer supports multiple backends through DreamerConfig.env_backend:

  • dmc: DeepMind Control Suite tasks (for example walker-walk)

  • gym: Gym/Gymnasium environment IDs or an existing environment instance

  • unity_mlagents: Unity ML-Agents executable environments

Important Unity settings are available in DreamerConfig:

  • unity_file_name

  • unity_behavior_name

  • unity_no_graphics

  • unity_time_scale

Typical Training Flow#

  1. Create a config object (DreamerConfig or JEPAConfig).

  2. Override dataset/environment and optimization fields.

  3. Instantiate the corresponding agent (DreamerAgent, JEPAAgent).

  4. Call train() and monitor logs/checkpoints.

For complete API details, see API Reference.