machina.pds package

Submodules

machina.pds.base module

class machina.pds.base.BasePd[source]

Bases: object

Base class of probablistic distribution

ent(params)[source]

entropy

kl_pq(p_params, q_params)[source]

KL divergence between p and q

llh(x, params)[source]

log liklihood

sample(params, sample_shape)[source]

sampling

machina.pds.categorical_pd module

Categorical

class machina.pds.categorical_pd.CategoricalPd[source]

Bases: machina.pds.base.BasePd

Categorical probablistic distribution.

ent(params)[source]

entropy

kl_pq(p_params, q_params)[source]

KL divergence between p and q

llh(x, params)[source]

log liklihood

sample(params, sample_shape=torch.Size([]))[source]

sampling

machina.pds.deterministic_pd module

class machina.pds.deterministic_pd.DeterministicPd[source]

Bases: machina.pds.base.BasePd

Deterministic probablistic distribution.

ent(params)[source]

entropy

kl_pq(p_params, q_params)[source]

KL divergence between p and q

llh(x, params)[source]

log liklihood

sample(params, sample_shape=torch.Size([]))[source]

sampling

machina.pds.gaussian_pd module

class machina.pds.gaussian_pd.GaussianPd[source]

Bases: machina.pds.base.BasePd

Gaussian probablistic distribution.

ent(params)[source]

entropy

kl_pq(p_params, q_params)[source]

KL divergence between p and q

llh(x, params)[source]

log liklihood

sample(params, sample_shape=torch.Size([]))[source]

sampling

machina.pds.mixture_gaussian_pd module

class machina.pds.mixture_gaussian_pd.MixtureGaussianPd(ob_space, ac_space)[source]

Bases: machina.pds.base.BasePd

kl_pq(p_params, q_params)[source]

KL divergence between p and q

llh(x, params)[source]

log liklihood

sample(params)[source]

sampling

machina.pds.multi_categorical_pd module

class machina.pds.multi_categorical_pd.MultiCategoricalPd[source]

Bases: machina.pds.base.BasePd

Multi Categorical probablistic distribution

ent(params)[source]

entropy

kl_pq(p_params, q_params)[source]

KL divergence between p and q

llh(xs, params)[source]

log liklihood

sample(params, sample_shape=torch.Size([]))[source]

sampling