`torch_mist.estimators.discriminative.base`

Module Contents

Classes

`DiscriminativeMIEstimator`
`BaselineDiscriminativeMIEstimator`

class torch_mist.estimators.discriminative.base.DiscriminativeMIEstimator(critic: torch_mist.critic.Critic, neg_samples: int = 1)

Bases: torch_mist.estimators.base.MIEstimator

lower_bound: bool = True

infomax_gradient: Dict[str, bool]

unnormalized_log_ratio(x: torch.Tensor, y: torch.Tensor) → torch.Tensor

n_negatives_to_use(N: int)

sample_negatives(x: torch.Tensor, y: torch.Tensor) → Tuple[torch.Tensor, torch.Tensor, torch.Tensor | None]

log_ratio(x: torch.Tensor, y: torch.Tensor) → torch.Tensor

abstract _approx_log_partition(x: torch.Tensor, y: torch.Tensor, f_: torch.Tensor, log_w: torch.Tensor | None) → torch.Tensor

approx_log_partition(x: torch.Tensor, y: torch.Tensor, x_: torch.Tensor, y_: torch.Tensor, log_w: torch.Tensor | None) → torch.Tensor

batch_loss(x: torch.Tensor, y: torch.Tensor) → torch.Tensor

__repr__()

class torch_mist.estimators.discriminative.base.BaselineDiscriminativeMIEstimator(critic: torch_mist.critic.Critic, baseline: torch_mist.baseline.Baseline, neg_samples: int = 1)

Bases: DiscriminativeMIEstimator

_approx_log_partition(x: torch.Tensor, y: torch.Tensor, f_: torch.Tensor, log_w: torch.Tensor | None) → torch.Tensor

__repr__()