Update to mypy 1.5

wearepal · Aug 11, 2023 · 1d405c6 · 1d405c6
1 parent dba342f
commit 1d405c6
Show file tree

Hide file tree

Showing 3 changed files with 11 additions and 10 deletions.
diff --git a/ethicml/implementations/fair_dummies_modules/model.py b/ethicml/implementations/fair_dummies_modules/model.py
@@ -492,7 +492,7 @@ def fit(self, train: DataTuple, seed: int) -> Self:
 
         for _ in range(self.pretrain_pred_epochs):
             random_array = rng.uniform(low=0.0, high=1.0, size=train.s.shape)
-            z_tilde = (random_array < p_success).astype(float)
+            z_tilde = (random_array < p_success).astype(np.dtype(float))
             zt_train = pd.DataFrame(data=z_tilde)
             train_data = PandasDataSet(train.x, train.s, train.y, zt_train)
             train_loader = DataLoader(
@@ -513,7 +513,7 @@ def fit(self, train: DataTuple, seed: int) -> Self:
 
         for _ in range(self.pretrain_dis_epochs):
             random_array = rng.uniform(low=0.0, high=1.0, size=train.s.shape)
-            z_tilde = (random_array < p_success).astype(float)
+            z_tilde = (random_array < p_success).astype(np.dtype(float))
             zt_train = pd.DataFrame(data=z_tilde)
             train_data = PandasDataSet(train.x, train.s, train.y, zt_train)
             train_loader = DataLoader(
@@ -536,7 +536,7 @@ def fit(self, train: DataTuple, seed: int) -> Self:
 
         for _ in range(1, self.epochs):
             random_array = rng.uniform(low=0.0, high=1.0, size=train.s.shape)
-            z_tilde = (random_array < p_success).astype(float)
+            z_tilde = (random_array < p_success).astype(np.dtype(float))
             zt_train = pd.DataFrame(data=z_tilde)
             train_data = PandasDataSet(train.x, train.s, train.y, zt_train)
             train_loader = DataLoader(
@@ -645,7 +645,7 @@ def fit(self, train: DataTuple, seed: int) -> None:
 
         for _ in range(self.pretrain_pred_epochs):
             random_array = rng.uniform(low=0.0, high=1.0, size=train.s.shape)
-            z_tilde = (random_array < p_success).astype(float)
+            z_tilde = (random_array < p_success).astype(np.dtype(float))
             zt_train = pd.DataFrame(data=z_tilde)
             train_data = PandasDataSet(train.x, train.s, train.y, zt_train)
             train_loader = DataLoader(
@@ -675,7 +675,7 @@ def fit(self, train: DataTuple, seed: int) -> None:
 
         for _ in range(self.pretrain_dis_epochs):
             random_array = rng.uniform(low=0.0, high=1.0, size=train.s.shape)
-            z_tilde = (random_array < p_success).astype(float)
+            z_tilde = (random_array < p_success).astype(np.dtype(float))
             zt_train = pd.DataFrame(data=z_tilde)
             train_data = PandasDataSet(train.x, train.s, train.y, zt_train)
             train_loader = DataLoader(
@@ -710,7 +710,7 @@ def fit(self, train: DataTuple, seed: int) -> None:
 
         for _ in range(1, self.epochs):
             random_array = rng.uniform(low=0.0, high=1.0, size=train.s.shape)
-            z_tilde = (random_array < p_success).astype(float)
+            z_tilde = (random_array < p_success).astype(np.dtype(float))
             zt_train = pd.DataFrame(data=z_tilde)
             train_data = PandasDataSet(train.x, train.s, train.y, zt_train)
             train_loader = DataLoader(

diff --git a/ethicml/models/inprocess/kamishima.py b/ethicml/models/inprocess/kamishima.py
@@ -84,7 +84,7 @@ def _predict(self, test: TestTuple, fit_info: _FitInfo, tmp_path: Path) -> Predi
         cmd = [script, "-i", test_path, "-m", fit_info.model_path, "-o", output_path, "--quiet"]
         self.call_script([str(e) for e in cmd])
         output = np.loadtxt(output_path)
-        predictions = output[:, 1].astype(np.float32)
+        predictions = output[:, 1].astype(np.dtype(np.float32))
         # except RuntimeError:
         #     predictions = np.ones_like(test.y.to_numpy())
 
@@ -102,9 +102,10 @@ def _create_file_in_kamishima_format(data: DataTuple | TestTuple, file_path: Pat
     :param data: Data to write to the file.
     :param file_path: Path to the file.
     """
+    dtype = np.dtype(np.float64)
     if isinstance(data, DataTuple):
-        result = pd.concat([data.x, data.s, data.y], axis="columns").to_numpy().astype(np.float64)
+        result = pd.concat([data.x, data.s, data.y], axis="columns").to_numpy().astype(dtype)
     else:
         zeros = pd.DataFrame([0 for _ in range(data.x.shape[0])], columns=["y"])
-        result = pd.concat([data.x, data.s, zeros], axis="columns").to_numpy().astype(np.float64)
+        result = pd.concat([data.x, data.s, zeros], axis="columns").to_numpy().astype(dtype)
     np.savetxt(file_path, result)
diff --git a/ethicml/models/postprocess/hardt.py b/ethicml/models/postprocess/hardt.py
@@ -183,7 +183,7 @@ def _predict(
         p2n_indices = othr_pp_indices[: int(len(othr_pp_indices) * (1 - op2p))]
         othr_fair_pred[p2n_indices] = self.unfavorable_label
 
-        new_labels = np.zeros_like(test_preds_numpy, dtype=np.float64)
+        new_labels = np.zeros_like(test_preds_numpy, dtype=np.dtype(np.float64))
         new_labels[mask_s1] = self_fair_pred
         new_labels[mask_s0] = othr_fair_pred