garage.misc.tensor_utils module

concat_tensor_dict_list(tensor_dict_list)[source]
concat_tensor_dict_list_subsample(tensor_dict_list, f)[source]
concat_tensor_list(tensor_list)[source]
concat_tensor_list_subsample(tensor_list, f)[source]
discount_cumsum(x, discount)[source]
explained_variance_1d(ypred, y)[source]
flatten_first_axis_tensor_dict(tensor_dict)[source]
flatten_tensors(tensors)[source]
high_res_normalize(probs)[source]
normalize_pixel_batch(env_spec, observations)[source]

Normalize the observations (images).

If the input are images, it normalized into range [0, 1].

Parameters:
  • env_spec (garage.envs.EnvSpec) – Environment specification.
  • observations (numpy.ndarray) – Observations from environment.
pad_tensor(x, max_len, mode='zero')[source]
pad_tensor_dict(tensor_dict, max_len, mode='zero')[source]
pad_tensor_n(xs, max_len)[source]
split_tensor_dict_list(tensor_dict)[source]
stack_tensor_dict_list(tensor_dict_list)[source]

Stack a list of dictionaries of {tensors or dictionary of tensors}. :param tensor_dict_list: a list of dictionaries of {tensors or dictionary

of tensors}.
Returns:a dictionary of {stacked tensors or dictionary of stacked tensors}
stack_tensor_list(tensor_list)[source]
truncate_tensor_dict(tensor_dict, truncated_len)[source]
truncate_tensor_list(tensor_list, truncated_len)[source]
unflatten_tensors(flattened, tensor_shapes)[source]