class flamo.optimize.dataset.Dataset(input: Tensor = tensor([[1.5410]]), target: Tensor = tensor([[-0.2934]]), expand: int = 1, device: str = 'cpu')

Base Dataset class. Inherits from torch.utils.data.Dataset.

In simple gradient descent optimization of differentiable DSP, both the input and target are represented by individual tensors. Therefore, the dataset consists of a single input tensor and a single target tensor. To accommodate larger datasets, the input and target tensors can be expanded to the desired dataset length using the expand attribute. This enables performing multiple optimization steps within a single epoch and allows for batch sizes greater than 1.

Arguments / Attributes:

  • input (torch.Tensor, optional): The input data tensor. Default: torch.randn(100, 100).

  • target (torch.Tensor, optional): The target data tensor. Default: torch.randn(100, 100).

  • expand (int): The first shape dimention of the input and target tensor after expansion. Default: 1. This coincides with the length on the dataset.

  • device (str, optional): The device to store the tensors on. Default: ‘cpu’.

class flamo.optimize.dataset.DatasetColorless(input_shape: tuple, target_shape: tuple, expand: int = 1000, device: str = 'cpu')

A dataset class for colorless optimization.

The input tensor is an impulse of length input_shape. The target is an instance of torch.ones with the shape target_shape, corresponding to a flat magnitude spectrum.

Arguments / Attributes:

  • input_shape (tuple): The shape of the input data.

  • target_shape (tuple): The shape of the target data.

  • expand (int, optional): The number of times to expand the dataset. Defaults to 1000.

  • device (str, optional): The device to use for computation. Defaults to ‘cpu’.

For details on the colorless optimization, see https://arxiv.org/abs/2402.11216v2.

flamo.optimize.dataset.get_dataloader(dataset: Dataset, batch_size: int = 2000, shuffle: bool = True)

Create a torch dataloader torch.utils.data.DataLoader from the given dataset.

Arguments:

  • dataset (torch.utils.data.Dataset): The dataset to create the dataloader from.

  • batch_size (int, optional): The number of samples per batch to load. Default: 2000.

  • shuffle (bool, optional): Whether to shuffle the dataset before each epoch. Default: True.

Returns:

torch.utils.data.DataLoader: The dataloader.

flamo.optimize.dataset.split_dataset(dataset: Dataset, split: float)

Splits the given dataset into a training set and a validation set based on the specified split ratio.

Arguments:

  • dataset (torch.utils.data.Dataset): The dataset to be split.

  • split (float): The ratio of the training set size to the total dataset size.

Returns:

tuple: A tuple containing the training set and the validation set.

flamo.optimize.dataset.load_dataset(dataset: Dataset, batch_size: int = 2000, split: float = 0.8, shuffle: bool = True)

Load and split a dataset into training and validation sets.

Arguments:

  • dataset (torch.utils.data.Dataset): The dataset to be loaded.

  • batch_size (int, optional): The batch size for the data loaders. Defaults to 2000.

  • split (float, optional): The ratio to split the dataset into training and validation sets. Defaults to 0.8.

  • shuffle (bool, optional): Whether to shuffle the data. Defaults to True.

Returns:

tuple: A tuple containing the training and validation data loaders.