ray.rllib.models.distributions.Distribution#

class ray.rllib.models.distributions.Distribution[source]#

Bases: ABC

The base class for distribution over a random variable.

Examples:

import torch
from ray.rllib.core.models.configs import MLPHeadConfig
from ray.rllib.models.torch.torch_distributions import TorchCategorical

model = MLPHeadConfig(input_dims=[1]).build(framework="torch")

# Create an action distribution from model logits
action_logits = model(torch.Tensor([[1]]))
action_dist = TorchCategorical.from_logits(action_logits)
action = action_dist.sample()

# Create another distribution from a dummy Tensor
action_dist2 = TorchCategorical.from_logits(torch.Tensor([0]))

# Compute some common metrics
logp = action_dist.logp(action)
kl = action_dist.kl(action_dist2)
entropy = action_dist.entropy()

Methods

`__init__`
`entropy`	The entropy of the distribution.
`from_logits`	Creates a Distribution from logits.
`get_partial_dist_cls`	Returns a partial child of TorchMultiActionDistribution.
`kl`	The KL-divergence between two distributions.
`logp`	The log-likelihood of the distribution computed at `value`
`required_input_dim`	Returns the required length of an input parameter tensor.
`rsample`	Draw a re-parameterized sample from the action distribution.
`sample`	Draw a sample from the distribution.
`to_deterministic`	Returns a deterministic equivalent for this distribution.