Language

unilab.algos.torch.him_ppo.storage¶

Classes

HIMRolloutStorage

class unilab.algos.torch.him_ppo.storage.HIMRolloutStorage[source]¶

Bases: object

Parameters:

num_envs (int)
num_transitions_per_env (int)
obs_shape (Sequence[int])
privileged_obs_shape (Sequence[int | None])
actions_shape (Sequence[int])
device (str)

class Transition[source]¶

Bases: object

__init__()[source]¶

clear()[source]¶

Return type:: None

__init__(num_envs, num_transitions_per_env, obs_shape, privileged_obs_shape, actions_shape, device='cpu')[source]¶

Parameters:

num_envs (int)
num_transitions_per_env (int)
obs_shape (Sequence[int])
privileged_obs_shape (Sequence[int | None])
actions_shape (Sequence[int])
device (str)

add_transition(transition)[source]¶

Parameters:: transition (Transition)
Return type:: None

add_transitions(transition)[source]¶

Parameters:: transition (Transition)
Return type:: None

clear()[source]¶

Return type:: None

compute_returns(last_values, gamma, lam)[source]¶

Parameters:

last_values (Tensor)
gamma (float)
lam (float)

Return type:

mini_batch_generator(num_mini_batches, num_epochs=8)[source]¶

Parameters:

num_mini_batches (int)
num_epochs (int)