Getting started

Install AMLGym:

pip install amlgym

Example usage for passive learning a domain model:

from amlgym.algorithms import get_algorithm
from amlgym.benchmarks import get_domain_path, get_trajectories_path
from amlgym.util.util import empty_domain

# Get a learning algorithm
agent = get_algorithm('offlam')

# Get an input domain path with no preconditions and effects
domain_path = get_domain_path('blocksworld')
domain_empty_path = empty_domain(domain_path)

# Get an input set of trajectory paths to learn from
traj_paths = get_trajectories_path('blocksworld')

# Learn a domain model
model = agent.learn(domain_empty_path, traj_paths)
print(model)

Further examples are available in the tutorials section for Learning Domain Models and Evaluating Domain Models.