Merge pull request #35 from aimagelab/dev

Big merge from dev

.github/workflows/autopep8.yml

@@ -0,0 +1,19 @@
+# This action works with pull requests and pushes
+name: Continuous Integration
+
+on:
+  pull_request:
+  push:
+    branches:
+    - master
+
+jobs:
+  build:
+    runs-on: ubuntu-latest
+
+    steps:
+    - name: autopep8
+      id: autopep8
+      uses: peter-evans/autopep8@v2
+      with:
+        args: --recursive --in-place --aggressive --max-line-length=200 --ignore=E402  .

.github/workflows/documentation.yml

@@ -0,0 +1,36 @@
+name: documentation
+
+on:
+  pull_request:
+    branches:
+      - release/wiki
+  push:
+    branches:
+      - release/wiki
+
+permissions:
+  contents: write
+
+jobs:
+  docs:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v3
+      - uses: actions/setup-python@v3
+        with:
+          python-version: "3.10"
+      - name: Install dependencies
+        run: |
+          pip install -r docs/requirements.txt -r requirements.txt
+          pip install quadprog==0.1.11
+      - name: Sphinx build
+        run: |
+          sphinx-build -j auto docs _build
+      - name: Deploy to GitHub Pages
+        uses: peaceiris/actions-gh-pages@v3
+        if: ${{ github.event_name == 'push' && github.ref == 'refs/heads/release/wiki' }}
+        with:
+          publish_branch: gh-pages
+          github_token: ${{ secrets.GITHUB_TOKEN }}
+          publish_dir: _build
+          force_orphan: true

.gitignore

@@ -2,6 +2,7 @@
 *.gif
 *.pdf
 data/
+data
 checkpoints/
 *.png
 *.pt
@@ -90,3 +91,8 @@ dmypy.json
 .pytype/
 cython_debug/
 wandb
+logs
+**/_build/
+_autosummary
+generated
+val_permutations

backbone/EfficientNet.py

@@ -155,9 +155,7 @@ def round_filters(filters, global_params):
     multiplier = global_params.width_coefficient
     if not multiplier:
         return filters
-    # TODO: modify the params names.
-    #       maybe the names (width_divisor,min_width)
-    #       are more suitable than (depth_divisor,min_depth).
+
     divisor = global_params.depth_divisor
     min_depth = global_params.min_depth
     filters *= multiplier
@@ -246,13 +244,6 @@ def get_same_padding_conv2d(image_size=None):
     else:
         return partial(Conv2dStaticSamePadding, image_size=image_size)
 
-# Parameters for the entire model (stem, all blocks, and head)
-# GlobalParams = collections.namedtuple('GlobalParams', [
-#     'width_coefficient', 'depth_coefficient', 'image_size', 'dropout_rate',
-#     'num_classes', 'batch_norm_momentum', 'batch_norm_epsilon',
-#     'drop_connect_rate', 'depth_divisor', 'min_depth', 'include_top'])
-
-
 GlobalParams = collections.namedtuple('GlobalParams', [
     'batch_norm_momentum', 'batch_norm_epsilon', 'dropout_rate', 'data_format',
     'num_classes', 'width_coefficient', 'depth_coefficient', 'depth_divisor',
@@ -261,11 +252,6 @@ def get_same_padding_conv2d(image_size=None):
     'blocks_args', 'image_size', 'drop_connect_rate', 'include_top'
 ])
 
-# Parameters for an individual model block
-# BlockArgs = collections.namedtuple('BlockArgs', [
-#     'num_repeat', 'kernel_size', 'stride', 'expand_ratio',
-#     'input_filters', 'output_filters', 'se_ratio', 'id_skip'])
-
 BlockArgs = collections.namedtuple('BlockArgs', [
     'kernel_size', 'num_repeat', 'input_filters', 'output_filters',
     'expand_ratio', 'id_skip', 'strides', 'se_ratio', 'conv_type', 'fused_conv',
@@ -965,47 +951,17 @@ def _change_in_channels(self, in_channels):
             out_channels = round_filters(32, self._global_params)
             self._conv_stem = Conv2d(in_channels, out_channels, kernel_size=3, stride=2, bias=False)
 
-    # TODO se va piallare
-    # def get_params(self, discard_classifier=False) -> torch.Tensor:
-    #     """
-    #     Returns all the parameters concatenated in a single tensor.
-    #     :return: parameters tensor (??)
-    #     """
-    #     params = []
-    #     for pp in list(self.parameters() if not discard_classifier else self._features.parameters()):
-    #         params.append(pp.view(-1))
-    #     return torch.cat(params)
-
-    # def set_params(self, new_params: torch.Tensor) -> None:
-    #     """
-    #     Sets the parameters to a given value.
-    #     :param new_params: concatenated values to be set (??)
-    #     """
-    #     assert new_params.size() == self.get_params().size()
-    #     progress = 0
-    #     for pp in list(self.parameters()):
-    #         cand_params = new_params[progress: progress +
-    #             torch.tensor(pp.size()).prod()].view(pp.size())
-    #         progress += torch.tensor(pp.size()).prod()
-    #         pp.data = cand_params
-
-    # def get_grads(self, discard_classifier=False) -> torch.Tensor:
-    #     """
-    #     Returns all the gradients concatenated in a single tensor.
-    #     :return: gradients tensor (??)
-    #     """
-    #     grads = []
-    #     for pp in list(self.parameters() if not discard_classifier else self._features.parameters()):
-    #         grads.append(pp.grad.view(-1))
-    #     return torch.cat(grads)
-
 
 def mammoth_efficientnet(nclasses: int, model_name: str, pretrained=False):
     """
     Instantiates a ResNet18 network.
-    :param nclasses: number of output classes
-    :param nf: number of filters
-    :return: ResNet network
+
+    Args:
+        nclasses: number of output classes
+        nf: number of filters
+
+    Returns:
+        ResNet network
     """
     print(model_name)
     if not pretrained:

backbone/MNISTMLP.py

@@ -18,8 +18,10 @@ class MNISTMLP(MammothBackbone):
     def __init__(self, input_size: int, output_size: int) -> None:
         """
         Instantiates the layers of the network.
-        :param input_size: the size of the input data
-        :param output_size: the size of the output
+
+        Args:
+            input_size: the size of the input data
+            output_size: the size of the output
         """
         super(MNISTMLP, self).__init__()
 
@@ -48,8 +50,12 @@ def reset_parameters(self) -> None:
     def forward(self, x: torch.Tensor, returnt='out') -> torch.Tensor:
         """
         Compute a forward pass.
-        :param x: input tensor (batch_size, input_size)
-        :return: output tensor (output_size)
+
+        Args:
+            x: input tensor (batch_size, input_size)
+
+        Returns:
+            output tensor (output_size)
         """
         x = x.view(-1, num_flat_features(x))
 
@@ -66,3 +72,8 @@ def forward(self, x: torch.Tensor, returnt='out') -> torch.Tensor:
             return (out, feats)
 
         raise NotImplementedError("Unknown return type")
+
+    def to(self, device):
+        super().to(device)
+        self.device = device
+        return self

backbone/MNISTMLP_PNN.py

@@ -23,9 +23,11 @@ def __init__(self, input_size: int, output_size: int,
                  old_cols: List[AlphaModule] = None) -> None:
         """
         Instantiates the layers of the network.
-        :param input_size: the size of the input data
-        :param output_size: the size of the output
-        :param old_cols: a list of all the old columns
+
+        Args:
+            input_size: the size of the input data
+            output_size: the size of the output
+            old_cols: a list of all the old columns
         """
         super(MNISTMLP_PNN, self).__init__()
 
@@ -84,8 +86,12 @@ def reset_parameters(self) -> None:
     def forward(self, x: torch.Tensor, returnt='out') -> torch.Tensor:
         """
         Compute a forward pass.
-        :param x: input tensor (batch_size, input_size)
-        :return: output tensor (output_size)
+
+        Args:
+            x: input tensor (batch_size, input_size)
+
+        Retruns:
+            output tensor (output_size)
         """
         x = x.view(-1, num_flat_features(x))
         if len(self.old_cols) > 0: