Module nn.loader.dataloader

Classes

class DataLoaders (x_train, y_train, batch_size, shuffle=True, x_val=None, y_val=None)

Expand source code

class DataLoaders:
    def __init__(
        self, x_train, y_train, batch_size, shuffle=True, x_val=None, y_val=None
    ):
        self.x_train = x_train
        self.y_train = y_train
        self.batch_size = batch_size
        self.shuffle = shuffle
        self.x_val = x_val
        self.y_val = y_val

    def flatten_images(self, images):
        # for now assuming, I am training images (with 3 dimensions)
        # so to get size, b x h (number of pixels per image)
        if images.ndim > 2:
            size = images.size(-1) * images.size(-2)
            return images.view(-1, size)
        return images

    def __len__(self):
        """Returns total number of data in single batch."""
        return len(self.x_train) // self.batch_size

    def __iter__(self):
        indices = np.arange(len(self.x_train))
        if self.shuffle:
            # perform shuffling
            np.random.shuffle(indices)

        # Implement minibatch iteration and return a list of mini-batches
        batches = []
        for i in range(0, len(self.x_train), self.batch_size):
            x_batch = self.flatten_images(
                self.x_train[indices[i : i + self.batch_size]]
            )
            y_batch = self.y_train[indices[i : i + self.batch_size]]
            batches.append((x_batch, y_batch))

        return iter(batches)

    def __str__(self):
        train_data = list(zip(self.x_train, self.y_train))
        return f"{train_data}"

    def get_validation_data(self):
        if self.x_val is not None and self.y_val is not None:
            return list(zip(self.x_val, self.y_val))
        return None

Methods

def flatten_images(self, images)

Expand source code

def flatten_images(self, images):
    # for now assuming, I am training images (with 3 dimensions)
    # so to get size, b x h (number of pixels per image)
    if images.ndim > 2:
        size = images.size(-1) * images.size(-2)
        return images.view(-1, size)
    return images

def get_validation_data(self)

Expand source code

def get_validation_data(self):
    if self.x_val is not None and self.y_val is not None:
        return list(zip(self.x_val, self.y_val))
    return None