module emote.memory.coverage_based_strategy
Classes
class CoverageBasedStrategy(Strategy):
A sampler intended to sample based on coverage of experiences, favoring less-visited states.
This base class can be used for implementing various coverage-based sampling strategies.
Methods
def __init__(self, alpha) -> None
Arguments:
alpha(default: 0.5)
def track(self, identity, sequence_length) -> None
def forget(self, identity) -> None
class CoverageBasedSampleStrategy(CoverageBasedStrategy, SampleStrategy):
Methods
def __init__(self, alpha) -> None
def sample(self, count, transition_count) -> Sequence[SamplePoint]