Merge branch 'main' into release

Author: tmke8

.github/workflows/continuous_integration.yml

@@ -30,39 +30,43 @@ jobs:
 
     steps:
     - uses: actions/checkout@v3
+    - name: Install poetry
+      run: pipx install poetry
     - name: Set up Python
       uses: actions/setup-python@v4
       with:
         python-version: '3.8'
     - name: Install ruff
       run: |
-        python -m pip install --upgrade pip
-        pip install ruff
+        poetry env use 3.8
+        poetry install --only lint --no-interaction --no-root
     - name: Lint with ruff
       run: |
-        ruff check --format=github ethicml
+        poetry run ruff check --format=github ethicml
     - name: Lint with ruff
       run: |
-        ruff check --format=github tests
+        poetry run ruff check --format=github tests
 
   format_with_black:
 
     runs-on: ubuntu-latest
 
     steps:
     - uses: actions/checkout@v3
+    - name: Install poetry
+      run: pipx install poetry
     - name: Set up Python
       uses: actions/setup-python@v4
       with:
         python-version: '3.8'
     - name: Install black
       run: |
-        python -m pip install --upgrade pip
-        pip install 'black<=23'
+        poetry env use 3.8
+        poetry install --only format --no-interaction --no-root
     - name: Format with black
       run: |
-        python -m black --check -l 100 -t py38 -S ethicml/
-        python -m black --check -l 100 -t py38 -S tests/
+        poetry run black --check -l 100 -t py38 -S ethicml/
+        poetry run black --check -l 100 -t py38 -S tests/
 
   check_docstrings:
 
@@ -135,7 +139,7 @@ jobs:
         run: |
           # keep the following in sync with `test_with_pytest`!
           poetry env use 3.8
-          poetry install --no-interaction --no-root -E all --with ci --with torchcpu
+          poetry install --no-interaction --no-root -E all --with ci --with torchcpu --without lint --without format
       - name: Test import
         run: |
           poetry run python -c "import ethicml"
@@ -166,16 +170,25 @@ jobs:
         run: |
           # keep the following in sync with `test_full_dependencies`!
           poetry env use 3.8
-          poetry install --no-interaction --no-root -E all --with ci --with torchcpu
+          poetry install --no-interaction --no-root -E all --with ci --with torchcpu --without lint --without format
 
       #----------------------------------------------
       #  -----           Run MyPy             -----
       #----------------------------------------------
       - name: Type check with mypy
         if: ${{ github.event_name == 'merge_group' }}
         run: |
-          poetry run python run_mypy.py
-          poetry run python run_mypy_tests.py
+          poetry run mypy ethicml
+          poetry run mypy --warn-unused-ignores ethicml || true
+
+      #----------------------------------------------
+      #  -----       Run MyPy on tests         -----
+      #----------------------------------------------
+      - name: Type check tests with mypy
+        if: ${{ github.event_name == 'merge_group' }}
+        run: |
+          poetry run mypy tests
+          poetry run mypy --warn-unused-ignores tests || true
 
       #----------------------------------------------
       #  -----           Run Tests            -----

.github/workflows/dependabot_auto.yml

@@ -1,8 +1,9 @@
 name: Dependabot auto-merge
 on:
   pull_request_target:
-    branches:    
-      - 'main'
+    branches:
+      # !! deactivated for now !!
+      - 'mainx'
 
 permissions:
   pull-requests: write

.github/workflows/docs.yml

@@ -1,36 +1,62 @@
-name: Build docs
+name: Deploy sphinx with GitHub Pages
 
 on:
   release:
     types:
       - published
 
+  # Allows you to run this workflow manually from the Actions tab
+  workflow_dispatch:
+
+# Sets permissions of the GITHUB_TOKEN to allow deployment to GitHub Pages
+permissions:
+  contents: read
+  pages: write
+  id-token: write
+
+# Allow only one concurrent deployment, skipping runs queued between the run in-progress and latest queued.
+# However, do NOT cancel in-progress runs as we want to allow these production deployments to complete.
+concurrency:
+  group: "pages"
+  cancel-in-progress: false
+
 jobs:
-  docs:
+  # Build job
+  build:
     runs-on: ubuntu-latest
     steps:
-    #----------------------------------------------
-    #       check-out repo and set-up python
-    #----------------------------------------------
-    - name: Check out repository
-      uses: actions/checkout@v2
-    - name: Set up python
-      uses: actions/setup-python@v2
-      with:
-        python-version: "3.10"
+      - name: Checkout
+        uses: actions/checkout@v3
+      - name: Setup Pages
+        uses: actions/configure-pages@v3
+      # --- build docs ---
+      - name: Set up python
+        uses: actions/setup-python@v4
+        with:
+          python-version: "3.10"
+      - name: Install sphinx
+        run: |
+          pip install wheel
+          pip install -r docs/requirements.txt
+      - name: Install pandoc
+        run: sudo apt-get install -y pandoc
+      - name: Build docs
+        run: |
+          sphinx-build -b html ./docs ./docs/_build
+      # -----------------
+      - name: Upload artifact
+        uses: actions/upload-pages-artifact@v2
+        with:
+          path: './docs/_build'
 
-    - name: Install sphinx
-      run: |
-        pip install wheel
-        pip install -r docs/requirements.txt
-    - name: Install pandoc
-      run: sudo apt-get install -y pandoc
-    - name: Build docs
-      run: |
-        sphinx-build -b html ./docs ./docs/_build
-
-    - name: Deploy
-      uses: peaceiris/actions-gh-pages@v3
-      with:
-        publish_dir: ./docs/_build
-        github_token: ${{ secrets.GITHUB_TOKEN }}
+  # Deployment job
+  deploy:
+    environment:
+      name: github-pages
+      url: ${{ steps.deployment.outputs.page_url }}
+    runs-on: ubuntu-latest
+    needs: build
+    steps:
+      - name: Deploy to GitHub Pages
+        id: deployment
+        uses: actions/deploy-pages@v2

ethicml/data/tabular_data/acs.py

@@ -6,7 +6,6 @@
 # pylint: skip-file
 # Pylint will go crazy as we're reimplementing the Dataset Init.
 from __future__ import annotations
-from collections.abc import Sequence
 import contextlib
 import os
 from pathlib import Path
@@ -322,7 +321,7 @@ def __init__(
         )
 
     @staticmethod
-    def cat_lookup(key: str) -> Sequence[int]:
+    def cat_lookup(key: str) -> list[int]:
         """Look up categories."""
         table = {
             "COW": range(1, 9),
@@ -333,7 +332,7 @@ def cat_lookup(key: str) -> Sequence[int]:
             'PINCP': range(2),
         }
 
-        return table[key]
+        return list(table[key])
 
     @override
     def load(
@@ -471,7 +470,7 @@ def __init__(
         )
 
     @staticmethod
-    def cat_lookup(key: str) -> Sequence[int]:
+    def cat_lookup(key: str) -> list[int]:
         """Look up categories."""
         table = {
             'SCHL': range(1, 25),
@@ -492,7 +491,7 @@ def cat_lookup(key: str) -> Sequence[int]:
             'ESR': range(1, 7),
         }
 
-        return table[key]
+        return list(table[key])
 
     @override
     def load(

ethicml/implementations/agarwal.py

@@ -123,15 +123,15 @@ def main() -> None:
         import cloudpickle  # pyright: ignore
 
         # Need to install cloudpickle for now. See https://github.com/fairlearn/fairlearn/issues/569
+
     except ImportError as e:
         raise RuntimeError("In order to use Agarwal, install fairlearn and cloudpickle.") from e
 
     if in_algo_args["mode"] == "run":
         random.seed(in_algo_args["seed"])
         np.random.seed(in_algo_args["seed"])
-        train, test = DataTuple.from_file(Path(in_algo_args["train"])), SubgroupTuple.from_file(
-            Path(in_algo_args["test"])
-        )
+        train = DataTuple.from_file(Path(in_algo_args["train"]))
+        test = SubgroupTuple.from_file(Path(in_algo_args["test"]))
         Prediction(
             hard=train_and_predict(train, test, flags, in_algo_args["seed"])["preds"]
         ).save_to_file(Path(in_algo_args["predictions"]))

ethicml/implementations/beutel.py

@@ -251,7 +251,7 @@ def backward(ctx: Any, grad_output: Tensor) -> Any:  # type: ignore[override]
 
 
 def _grad_reverse(features: Tensor, lambda_: float) -> Tensor:
-    return GradReverse.apply(features, lambda_)
+    return GradReverse.apply(features, lambda_)  # pyright: ignore
 
 
 class Encoder(nn.Module):

ethicml/implementations/dro_modules/dro_classifier.py

@@ -16,7 +16,7 @@ class DROClassifier(nn.Module):
     def __init__(self, in_size: int, out_size: int, network_size: list[int], eta: float) -> None:
         super().__init__()
 
-        self.loss = DROLoss(loss_module=CrossEntropyLoss, eta=eta)
+        self.loss = DROLoss(loss_module=CrossEntropyLoss, eta=eta)  # type: ignore[arg-type]
 
         self.seq = nn.Sequential()
         if not network_size:  # In the case that encoder size [] is specified

ethicml/implementations/dro_tabular.py

@@ -47,13 +47,15 @@ def train_model(
         optimizer.step()
         train_loss += loss.item()
         if batch_idx % 100 == 0:
+            num_epochs = len(train_loader.dataset)  # type: ignore[arg-type]
             print(
-                f"train Epoch: {epoch} [{batch_idx * len(data_x)}/{len(train_loader.dataset)}"
+                f"train Epoch: {epoch} [{batch_idx * len(data_x)}/{num_epochs}"
                 f"\t({100. * batch_idx / len(train_loader):.0f}%)]"
                 f"\tLoss: {loss.item() / len(data_x):.6f}"
             )
 
-    print(f"====> Epoch: {epoch} Average loss: {train_loss / len(train_loader.dataset):.4f}")
+    num_epochs = len(train_loader.dataset)  # type: ignore[arg-type]
+    print(f"====> Epoch: {epoch} Average loss: {train_loss / num_epochs:.4f}")
 
 
 def fit(train: DataTuple, args: DroArgs, seed: int) -> DROClassifier:

ethicml/implementations/fair_dummies_modules/model.py

@@ -483,7 +483,7 @@ def __init__(
     def fit(self, train: DataTuple, seed: int) -> Self:
         """Fit."""
         # The features are X[:,1:]
-        p_success, dummy = density_estimation(y=train.y.to_numpy(), a=train.s.to_numpy())
+        p_success, _ = density_estimation(y=train.y.to_numpy(), a=train.s.to_numpy())
 
         rng = np.random.default_rng(seed)
 
@@ -633,7 +633,7 @@ def __init__(
     def fit(self, train: DataTuple, seed: int) -> None:
         """Fit."""
         fast_loader = self.batch_size >= train.x.shape[0]
-        p_success, dummy = density_estimation(y=train.y.to_numpy(), a=train.s.to_numpy())
+        p_success, _ = density_estimation(y=train.y.to_numpy(), a=train.s.to_numpy())
 
         x = torch.from_numpy(train.x.to_numpy()).float()
         y = torch.from_numpy(train.y.to_numpy()).float()

ethicml/implementations/fair_dummies_modules/utility_functions.py

@@ -12,7 +12,7 @@ def density_estimation(
     if y_test is None:
         y_test = np.array([])
     assert y_test is not None
-    bandwidth = np.sqrt(max(np.median(np.abs(y)), 0.01))
+    bandwidth = np.sqrt(max(np.median(np.abs(y)).item(), 0.01))
 
     kde_0 = KernelDensity(kernel='linear', bandwidth=bandwidth).fit(y[a == 0][:, np.newaxis])
     kde_1 = KernelDensity(kernel='linear', bandwidth=bandwidth).fit(y[a == 1][:, np.newaxis])

ethicml/implementations/fair_dummies_romano.py

@@ -63,9 +63,8 @@ def main() -> None:
     flags: FairDummiesArgs = json.loads(sys.argv[2])
 
     if in_algo_args["mode"] == "run":
-        train, test = DataTuple.from_file(Path(in_algo_args["train"])), SubgroupTuple.from_file(
-            Path(in_algo_args["test"])
-        )
+        train = DataTuple.from_file(Path(in_algo_args["train"]))
+        test = SubgroupTuple.from_file(Path(in_algo_args["test"]))
         SoftPrediction(
             soft=train_and_predict(train, test, flags, in_algo_args["seed"])
         ).save_to_file(Path(in_algo_args["predictions"]))

ethicml/implementations/hgr_method.py

@@ -65,9 +65,8 @@ def main() -> None:
     flags: HgrArgs = json.loads(sys.argv[2])
 
     if in_algo_args["mode"] == "run":
-        train, test = DataTuple.from_file(Path(in_algo_args["train"])), SubgroupTuple.from_file(
-            Path(in_algo_args["test"])
-        )
+        train = DataTuple.from_file(Path(in_algo_args["train"]))
+        test = SubgroupTuple.from_file(Path(in_algo_args["test"]))
         SoftPrediction(
             soft=train_and_predict(train, test, flags, in_algo_args["seed"])
         ).save_to_file(Path(in_algo_args["predictions"]))

ethicml/implementations/hgr_modules/facl_hgr.py

@@ -49,7 +49,7 @@ def hgr(x: torch.Tensor, y: torch.Tensor, density: Type[Kde], damping: float = 1
     return torch.svd(Q)[1][1]
 
 
-def chi_2(x: torch.Tensor, y: torch.Tensor, density: Type[Kde], damping: float = 0) -> float:
+def chi_2(x: torch.Tensor, y: torch.Tensor, density: Type[Kde], damping: float = 0) -> torch.Tensor:
     r"""The :math:`\chi^2` divergence between the joint distribution and the product of marginals.
 
     This is know to be the square of an upper-bound on the Hirschfeld-Gebelein-Renyi maximum