jaclearn.rl.algo package

Submodules

jaclearn.rl.algo.advantage module

class jaclearn.rl.algo.advantage.AdvantageComputerBase

Bases: object

class jaclearn.rl.algo.advantage.DiscountedAdvantageComputer(gamma)

Bases: jaclearn.rl.algo.advantage.AdvantageComputerBase

class jaclearn.rl.algo.advantage.GAEComputer(gamma, lambda_)

Bases: jaclearn.rl.algo.advantage.AdvantageComputerBase

jaclearn.rl.algo.math module

class jaclearn.rl.algo.math.LinearValueRegressor

Bases: object

coeffs = None

fit(states, steps, returns)

predict(states, steps)

register_snapshot_parts(env)

class jaclearn.rl.algo.math.ObservationNormalizer(filter_mean=True)

Bases: object

normalize(o)

jaclearn.rl.algo.math.compute_gae(rewards, values, next_val, gamma, lambda_)

jaclearn.rl.algo.math.discount_cumsum(x, gamma)

Compute the discounted cumulative summation of an 1-d array. From https://github.com/rll/rllab/blob/master/rllab/misc/special.py

jaclearn.rl.algo.math.discount_return(x, discount)

Compute the discounted return summation of an 1-d array. From https://github.com/rll/rllab/blob/master/rllab/misc/special.py

jaclearn.rl.algo.math.normalize_advantage(adv)