Samples actions from the policy distribution. Args: obs_dict (dict): batch of observations goal_dict (dict):

Added kwargs vs diffuser's original implementation Unified API to get any scheduler from its name. Args: name (`str` or `Schedu

Record data with logger. Args: k (str): key string v (float or image): value to store epoch: curr

predicate: Return true if the module is to be replaced. func: Return new module to use.

Args: obs_shapes (OrderedDict): a dictionary that maps modality to expected shapes for observations.

Factory function for creating algorithms based on the algorithm name and config. Args: algo_name (str): the algorithm name

Calls the method with name and applies args and kwargs. Args: name: The method name *args: The method args

Create a SharedNDArray directly from a SharedMemoryManager Parameters ---------- mem_mgr SharedMemoryManager insta

Args: value (torch.Tensor): some tensor to compute log probabilities for pre_tanh_value: If specified, will not calcu

Function to compute output shape from inputs to this module. Args: input_shape (iterable of int): shape of input. Does n

Reset to a specific simulator state. Args: state (dict): current simulator state that contains: - states

This contextmanager will redirect stdout so that nothing is printed to the terminal. Taken from the link below: https://stackoverflow.co

Add parameter to the hyperparameter sweep. Args: key (str): location of parameter in the config, using hierarchical key

Stop recording

For each dt window, choose the first timestamp in the window. Assumes timestamps sorted. One timestamp might be chosen multiple times due to

observation dict

....

Prepares an observation @obs of this modality for network input. Args: obs (np.array or torch.Tensor): raw observation,

Returns environment type (int) for this kind of environment. This helps identify this env class.

Uses registry to retrieve algo factory function from algo name. Args: algo_name (str): the algorithm name

Input: points: input points data, [B, N, C] idx: sample index data, [B, S] Return: new_points:, indexed points data,

Apply function @func to torch.Tensor objects in a nested dictionary or list or tuple. Args: x (dict or list or tuple): a possibl

Apply function @func to torch.Tensor objects in a nested dictionary or list or tuple. Args: x (dict or list or tuple): a possibl

Processes input batch from a data loader to filter out relevant information and prepare the batch for training. Args:

Render from simulation to either an on-screen window or off-screen to RGB array. Args: mode (str): pass "human" for on-s

Reset algo state to prepare for environment rollouts.

Args: input_obs_group_shapes (OrderedDict): a dictionary of dictionaries. Each key in this dictionary should spec

A full pass through the VAE network to construct KL and reconstruction losses. Args: inputs (dict): a dictionary

Helper function to generate a config for a particular experiment. Args: base_exp_name (str): name that identifies this set of experi

Extract a (sub)sequence of observation items from a demo given the @keys of the items. Args: demo_id (str): id of the de