Language

unilab.algos.mlx.ppo.model.MLPActorCritic¶

class unilab.algos.mlx.ppo.model.MLPActorCritic[source]¶

Bases: Module

Shared utility class containing actor and critic MLPs.

Parameters:

obs_dim (int)
action_dim (int)
actor_hidden_dims (Sequence[int])
critic_hidden_dims (Sequence[int])
activation (str)
init_log_std (float)
min_log_std (float)
max_log_std (float)
obs_normalization (bool)
noise_std_type (str)
state_dependent_std (bool)
dtype (Any | None)

Methods

`__init__`(obs_dim, action_dim, ...[, ...])
`act`(obs)	Sample actions and return MLX tensors.
`clipped_log_std`()	Clamp log-std to avoid numerical explosion.
`current_action_std`(action_shape)	Return broadcasted std tensor for current policy.
`distribution_params`(obs)
`policy`(obs)
`update_normalization`(obs)
`value`(obs)

__init__(obs_dim, action_dim, actor_hidden_dims, critic_hidden_dims, activation='tanh', init_log_std=0.0, min_log_std=-5.0, max_log_std=2.0, obs_normalization=False, noise_std_type='log', state_dependent_std=False, dtype=None)[source]¶

Parameters:

obs_dim (int)
action_dim (int)
actor_hidden_dims (Sequence[int])
critic_hidden_dims (Sequence[int])
activation (str)
init_log_std (float)
min_log_std (float)
max_log_std (float)
obs_normalization (bool)
noise_std_type (str)
state_dependent_std (bool)
dtype (Any | None)

clipped_log_std()[source]¶

Clamp log-std to avoid numerical explosion.

Return type:: array

policy(obs)[source]¶

Parameters:: obs (array)
Return type:: array

distribution_params(obs)[source]¶

Parameters:: obs (array)
Return type:: tuple[array, array, array]

value(obs)[source]¶

Parameters:: obs (array)
Return type:: array

update_normalization(obs)[source]¶

Parameters:: obs (array)
Return type:: None

act(obs)[source]¶

Sample actions and return MLX tensors.

Parameters:: obs (array)
Return type:: Tuple[array, array, array, array, array]

current_action_std(action_shape)[source]¶

Return broadcasted std tensor for current policy.

Parameters:: action_shape (tuple[int, ...])
Return type:: array