Reinforcement Learning¶

Module contents¶

class olympus.reinforcement.Environment(env_name, transforms=None, rand_seed=None, train_size=1024, valid_size=128, test_size=128, parallel_env=8, num_thread=4, distribution_mode='easy')[source]¶

Bases: object

Generic Reinforcement Learning Environment, can run multiple simulation in parallel

Attributes:	action_space `input_size` Return the size of the samples state_space `target_size` Return the size of the target test train valid

Methods

categories() Dataset tags so we can filter what we want depending on the task

close
load_state_dict
max
sample_action
state_dict

action_space¶

categories()[source]¶: Dataset tags so we can filter what we want depending on the task

close()[source]¶

input_size¶: Return the size of the samples

load_state_dict(data)[source]¶

max()[source]¶

sample_action()[source]¶

state_dict()[source]¶

state_space¶

target_size¶: Return the size of the target

test¶

train¶

valid¶

exception olympus.reinforcement.RegisteredEnvironmentNotFound[source]¶: Bases: Exception

olympus.reinforcement.known_environments(*category_filters, include_unknown=False)[source]¶: List known environments

olympus.reinforcement.register_environment(name, factory, override=False)[source]¶: Register a new environment backend