prl package

Submodules

prl.typing module

class ActionTransformerABC[source]

Bases: abc.ABC

action_space(original_space)[source]
Return type:Space
id
Return type:str
reset()[source]
transform(action, history)[source]
Return type:ndarray
class AdvantageABC[source]

Bases: abc.ABC

class AgentABC[source]

Bases: abc.ABC

act(state)[source]
id
Return type:str
play_episodes(env, episodes)[source]
Return type:HistoryABC
play_steps(env, n_steps, history)[source]
Return type:HistoryABC
post_train_cleanup(env, **kwargs)[source]
pre_train_setup(env, **kwargs)[source]
test(env)[source]
Return type:HistoryABC
train(env, n_iterations, callback_list, **kwargs)[source]
train_iteration(env, **kwargs)[source]
Return type:(<class ‘float’>, <class ‘prl.typing.HistoryABC’>)
class AgentCallbackABC[source]

Bases: abc.ABC

on_iteration_end(agent)[source]
Return type:bool
on_training_begin(agent)[source]
on_training_end(agent)[source]
class EnvironmentABC[source]

Bases: abc.ABC

action_space
Return type:Space
action_transformer
Return type:ActionTransformerABC
close()[source]
id
observation_space
Return type:Space
reset()[source]
Return type:ndarray
reward_transformer
Return type:RewardTransformerABC
state_history
Return type:HistoryABC
state_transformer
Return type:StateTransformerABC
step(action)[source]
Return type:Tuple[ndarray, Real, bool, Dict[~KT, ~VT]]
class FunctionApproximatorABC[source]

Bases: abc.ABC

id
Return type:str
predict(x)[source]
train(x, *loss_args)[source]
Return type:float
class HistoryABC[source]

Bases: abc.ABC

get_actions()[source]
Return type:ndarray
get_dones()[source]
Return type:ndarray
get_last_state()[source]
Return type:ndarray
get_number_of_episodes()[source]
Return type:int
get_returns(discount_factor, horizon)[source]
Return type:ndarray
get_rewards()[source]
Return type:ndarray
get_states()[source]
Return type:ndarray
get_summary()[source]
get_total_rewards()[source]
Return type:ndarray
new_state_update(state)[source]
sample_batch(replay_buffor_size, batch_size, returns, next_states)[source]
Return type:tuple
update(action, reward, done, state)[source]
MemoryABC

alias of prl.typing.StorageABC

class PytorchNetABC(*args, **kwargs)[source]

Bases: sphinx.ext.autodoc.importer._MockObject

forward(x)[source]
predict(x)[source]
class RewardTransformerABC[source]

Bases: abc.ABC

id
Return type:str
reset()[source]
transform(reward, history)[source]
Return type:Real
class StateTransformerABC[source]

Bases: abc.ABC

id
Return type:str
reset()[source]
transform(state, history)[source]
Return type:ndarray
class StorageABC[source]

Bases: abc.ABC

get_actions()[source]
Return type:ndarray
get_dones()[source]
Return type:ndarray
get_last_state()[source]
Return type:ndarray
get_rewards()[source]
Return type:ndarray
get_states()[source]
Return type:ndarray
new_state_update(state)[source]
sample_batch(replay_buffor_size, batch_size, returns, next_states)[source]
Return type:tuple
update(action, reward, done, state)[source]

Module contents