# -*- coding: utf-8 -*-
"""A vanilla MLP architecture for MNIST."""
from torch import nn
from .testproblems_modules import net_mlp
from ..datasets.mnist import mnist
from .testproblem import UnregularizedTestproblem
import warnings
[docs]class mnist_mlp(UnregularizedTestproblem):
"""DeepOBS test problem class for a multi-layer perceptron neural network\
on Fashion-MNIST.
The network is build as follows:
- Four fully-connected layers with ``1000``, ``500``, ``100`` and ``10``
units per layer.
- The first three layers use ReLU activation, and the last one a softmax
activation.
- The biases are initialized to ``0.0`` and the weight matrices with
truncated normal (standard deviation of ``3e-2``)
- The model uses a cross entropy loss.
- No regularization is used.
Args:
batch_size (int): Batch size to use.
weight_decay (float): No weight decay (L2-regularization) is used in this
test problem. Defaults to ``None`` and any input here is ignored.
Attributes:
data: The DeepOBS data set class for Fashion-MNIST.
loss_function: The loss function for this testproblem is torch.nn.CrossEntropyLoss()
net: The DeepOBS subclass of torch.nn.Module that is trained for this tesproblem (net_mlp).
"""
def __init__(self, batch_size, weight_decay=None):
"""Create a new multi-layer perceptron test problem instance on \
Fashion-MNIST.
Args:
batch_size (int): Batch size to use.
weight_decay (float): No weight decay (L2-regularization) is used in this
test problem. Defaults to ``None`` and any input here is ignored.
"""
super(mnist_mlp, self).__init__(batch_size, weight_decay)
if weight_decay is not None:
warnings.warn(
"Weight decay is non-zero but no weight decay is used for this model.",
RuntimeWarning
)
[docs] def set_up(self):
"""Sets up the vanilla CNN test problem on MNIST."""
self.data = mnist(self._batch_size)
self.loss_function = nn.CrossEntropyLoss
self.net = net_mlp(num_outputs=10)
self.net.to(self._device)
self.regularization_groups = self.get_regularization_groups()