diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml
index cbe6be56..fc254ba9 100644
--- a/.pre-commit-config.yaml
+++ b/.pre-commit-config.yaml
@@ -27,8 +27,8 @@ repos:
       # mypy
       - id: mypy
         name: mypy
-        entry: pixi run -e default mypy
-        exclude: (^tests/|^src/glum_benchmarks/orig_sklearn_fork/)
+        entry: pixi run -e default mypy --allow-redefinition
+        exclude: (^tests/)
         language: system
         types: [python]
         require_serial: true
diff --git a/src/tabmat/categorical_matrix.py b/src/tabmat/categorical_matrix.py
index 33269439..60bc53d5 100644
--- a/src/tabmat/categorical_matrix.py
+++ b/src/tabmat/categorical_matrix.py
@@ -252,7 +252,9 @@ def _extract_codes_and_categories(cat_vec):
 
 
 def _row_col_indexing(
-    arr: np.ndarray, rows: Optional[np.ndarray], cols: Optional[np.ndarray]
+    arr: Union[np.ndarray, sps.spmatrix],
+    rows: Optional[np.ndarray],
+    cols: Optional[np.ndarray],
 ) -> np.ndarray:
     if isinstance(rows, slice) and rows == slice(None, None, None):
         rows = None
@@ -411,7 +413,7 @@ def recover_orig(self) -> np.ndarray:
     def _matvec_setup(
         self,
         other: Union[list, np.ndarray],
-        cols: np.ndarray = None,
+        cols: Optional[np.ndarray] = None,
     ) -> tuple[np.ndarray, Optional[np.ndarray]]:
         other = np.asarray(other)
         if other.ndim > 1:
@@ -434,8 +436,8 @@ def _matvec_setup(
     def matvec(
         self,
         other: Union[list, np.ndarray],
-        cols: np.ndarray = None,
-        out: np.ndarray = None,
+        cols: Optional[np.ndarray] = None,
+        out: Optional[np.ndarray] = None,
     ) -> np.ndarray:
         """
         Multiply self with vector 'other', and add vector 'out' if it is present.
@@ -524,8 +526,7 @@ def transpose_matvec(
                 "CategoricalMatrix.transpose_matvec is only implemented for 1d arrays."
             )
 
-        out_is_none = out is None
-        if out_is_none:
+        if out_is_none := out is None:
             out = np.zeros(self.shape[1], dtype=self.dtype)
         else:
             check_transpose_matvec_out_shape(self, out)
@@ -558,8 +559,8 @@ def transpose_matvec(
     def sandwich(
         self,
         d: Union[np.ndarray, list],
-        rows: np.ndarray = None,
-        cols: np.ndarray = None,
+        rows: Optional[np.ndarray] = None,
+        cols: Optional[np.ndarray] = None,
     ) -> sps.dia_matrix:
         """
         Perform a sandwich product: X.T @ diag(d) @ X.
@@ -594,7 +595,7 @@ def sandwich(
     def _cross_sandwich(
         self,
         other: MatrixBase,
-        d: Union[np.ndarray, list],
+        d: np.ndarray,
         rows: Optional[np.ndarray] = None,
         L_cols: Optional[np.ndarray] = None,
         R_cols: Optional[np.ndarray] = None,
diff --git a/src/tabmat/constructor.py b/src/tabmat/constructor.py
index b1c73bc7..2c33130b 100644
--- a/src/tabmat/constructor.py
+++ b/src/tabmat/constructor.py
@@ -21,11 +21,11 @@
 try:
     import polars as pl
 except ImportError:
-    pl = None
+    pl = None  # type: ignore
 try:
     import pandas as pd
 except ImportError:
-    pd = None
+    pd = None  # type: ignore
 
 
 def _is_boolean(series, engine: str):
@@ -106,7 +106,7 @@ def _from_dataframe(
     """
 
     matrices: list[Union[DenseMatrix, SparseMatrix, CategoricalMatrix]] = []
-    indices: list[list[int]] = []
+    indices: list[np.ndarray] = []
     is_cat: list[bool] = []
 
     dense_dfidx = []  # column index in original DataFrame
@@ -197,7 +197,7 @@ def _from_dataframe(
                 term_names=np.asarray(df.columns)[dense_dfidx],
             )
         )
-        indices.append(dense_tmidx)
+        indices.append(np.asarray(dense_tmidx))
         is_cat.append(False)
     if sparse_dfidx:
         matrices.append(
@@ -208,7 +208,7 @@ def _from_dataframe(
                 term_names=np.asarray(df.columns)[sparse_dfidx],
             )
         )
-        indices.append(sparse_tmidx)
+        indices.append(np.asarray(sparse_tmidx))
         is_cat.append(False)
 
     if cat_position == "end":
@@ -357,7 +357,7 @@ def from_polars(
     )
 
 
-def _reindex_cat(indices, is_cat, mxcolidx):
+def _reindex_cat(indices: list[np.ndarray], is_cat: list[bool], mxcolidx: int):
     new_indices = []
     for mat_indices, is_cat_ in zip(indices, is_cat):
         if is_cat_:
diff --git a/src/tabmat/dense_matrix.py b/src/tabmat/dense_matrix.py
index 0dd27845..3eca4fbc 100644
--- a/src/tabmat/dense_matrix.py
+++ b/src/tabmat/dense_matrix.py
@@ -136,7 +136,10 @@ def unpack(self):
         return self._array
 
     def sandwich(
-        self, d: np.ndarray, rows: np.ndarray = None, cols: np.ndarray = None
+        self,
+        d: np.ndarray,
+        rows: Optional[np.ndarray] = None,
+        cols: Optional[np.ndarray] = None,
     ) -> np.ndarray:
         """Perform a sandwich product: X.T @ diag(d) @ X."""
         d = np.asarray(d)
@@ -219,9 +222,9 @@ def _matvec_helper(
     def transpose_matvec(
         self,
         vec: Union[np.ndarray, list],
-        rows: np.ndarray = None,
-        cols: np.ndarray = None,
-        out: np.ndarray = None,
+        rows: Optional[np.ndarray] = None,
+        cols: Optional[np.ndarray] = None,
+        out: Optional[np.ndarray] = None,
     ) -> np.ndarray:
         """Perform: self[rows, cols].T @ vec[rows]."""
         check_transpose_matvec_out_shape(self, out)
@@ -230,8 +233,8 @@ def transpose_matvec(
     def matvec(
         self,
         vec: Union[np.ndarray, list],
-        cols: np.ndarray = None,
-        out: np.ndarray = None,
+        cols: Optional[np.ndarray] = None,
+        out: Optional[np.ndarray] = None,
     ) -> np.ndarray:
         """Perform self[:, cols] @ other[cols]."""
         check_matvec_out_shape(self, out)
diff --git a/src/tabmat/matrix_base.py b/src/tabmat/matrix_base.py
index b1236f52..3670a08c 100644
--- a/src/tabmat/matrix_base.py
+++ b/src/tabmat/matrix_base.py
@@ -12,7 +12,9 @@ class MatrixBase(ABC):
     dtype: np.dtype
 
     @abstractmethod
-    def matvec(self, other, cols: np.ndarray = None, out: np.ndarray = None):
+    def matvec(
+        self, other, cols: Optional[np.ndarray] = None, out: Optional[np.ndarray] = None
+    ):
         """
         Perform: self[:, cols] @ other[cols], so result[i] = sum_j self[i, j] other[j].
 
@@ -32,9 +34,9 @@ def matvec(self, other, cols: np.ndarray = None, out: np.ndarray = None):
     def transpose_matvec(
         self,
         vec: Union[np.ndarray, list],
-        rows: np.ndarray = None,
-        cols: np.ndarray = None,
-        out: np.ndarray = None,
+        rows: Optional[np.ndarray] = None,
+        cols: Optional[np.ndarray] = None,
+        out: Optional[np.ndarray] = None,
     ) -> np.ndarray:
         """
         Perform: self[rows, cols].T @ vec[rows], so result[i] = sum_j self[j, i] vec[j].
@@ -61,7 +63,10 @@ def transpose_matvec(
 
     @abstractmethod
     def sandwich(
-        self, d: np.ndarray, rows: np.ndarray = None, cols: np.ndarray = None
+        self,
+        d: np.ndarray,
+        rows: Optional[np.ndarray] = None,
+        cols: Optional[np.ndarray] = None,
     ) -> np.ndarray:
         """
         Perform a sandwich product: (self[rows, cols].T * d[rows]) @ self[rows, cols].
diff --git a/src/tabmat/sparse_matrix.py b/src/tabmat/sparse_matrix.py
index 256f7caa..5182becb 100644
--- a/src/tabmat/sparse_matrix.py
+++ b/src/tabmat/sparse_matrix.py
@@ -172,7 +172,10 @@ def dot(self, other):
         return self._array.dot(other)
 
     def sandwich(
-        self, d: np.ndarray, rows: np.ndarray = None, cols: np.ndarray = None
+        self,
+        d: np.ndarray,
+        rows: Optional[np.ndarray] = None,
+        cols: Optional[np.ndarray] = None,
     ) -> np.ndarray:
         """Perform a sandwich product: X.T @ diag(d) @ X."""
         d = np.asarray(d)
@@ -190,7 +193,7 @@ def _cross_sandwich(
         self,
         other: MatrixBase,
         d: np.ndarray,
-        rows: np.ndarray,
+        rows: Optional[np.ndarray],
         L_cols: Optional[np.ndarray] = None,
         R_cols: Optional[np.ndarray] = None,
     ):
@@ -209,9 +212,9 @@ def sandwich_dense(
         self,
         B: np.ndarray,
         d: np.ndarray,
-        rows: np.ndarray,
-        L_cols: np.ndarray,
-        R_cols: np.ndarray,
+        rows: Optional[np.ndarray],
+        L_cols: Optional[np.ndarray],
+        R_cols: Optional[np.ndarray],
     ) -> np.ndarray:
         """Perform a sandwich product: self.T @ diag(d) @ B."""
         if not hasattr(d, "dtype"):
@@ -276,7 +279,9 @@ def _matvec_helper(
             out[rows] += res
         return out
 
-    def matvec(self, vec, cols: np.ndarray = None, out: np.ndarray = None):
+    def matvec(
+        self, vec, cols: Optional[np.ndarray] = None, out: Optional[np.ndarray] = None
+    ):
         """Perform self[:, cols] @ other[cols]."""
         check_matvec_out_shape(self, out)
         return self._matvec_helper(vec, None, cols, out, False)
@@ -284,9 +289,9 @@ def matvec(self, vec, cols: np.ndarray = None, out: np.ndarray = None):
     def transpose_matvec(
         self,
         vec: Union[np.ndarray, list],
-        rows: np.ndarray = None,
-        cols: np.ndarray = None,
-        out: np.ndarray = None,
+        rows: Optional[np.ndarray] = None,
+        cols: Optional[np.ndarray] = None,
+        out: Optional[np.ndarray] = None,
     ) -> np.ndarray:
         """Perform: self[rows, cols].T @ vec[rows]."""
         check_transpose_matvec_out_shape(self, out)
diff --git a/src/tabmat/split_matrix.py b/src/tabmat/split_matrix.py
index 6281c5c5..6f44f73a 100644
--- a/src/tabmat/split_matrix.py
+++ b/src/tabmat/split_matrix.py
@@ -28,7 +28,7 @@ def as_tabmat(a: Union[MatrixBase, StandardizedMatrix, np.ndarray, sps.spmatrix]
     if isinstance(a, (MatrixBase, StandardizedMatrix)):
         return a
     elif sps.issparse(a):
-        return SparseMatrix(a.tocsc(copy=False))
+        return SparseMatrix(a.tocsc(copy=False))  # type: ignore
     elif isinstance(a, np.ndarray):
         return DenseMatrix(a)
     else:
@@ -59,7 +59,7 @@ def hstack(tup: Sequence[Union[MatrixBase, np.ndarray, sps.spmatrix]]) -> Matrix
         return SplitMatrix(matrices)
 
 
-def _prepare_out_array(out: Optional[np.ndarray], out_shape, out_dtype):
+def _prepare_out_array(out: Optional[np.ndarray], out_shape, out_dtype) -> np.ndarray:
     if out is None:
         out = np.zeros(out_shape, out_dtype)
     else:
@@ -266,7 +266,7 @@ def __init__(
 
     def _split_col_subsets(
         self, cols: Optional[np.ndarray]
-    ) -> tuple[list[np.ndarray], list[Optional[np.ndarray]], int]:
+    ) -> tuple[list[np.ndarray], Union[list[np.ndarray], list[None]], int]:
         """
         Return tuple of things helpful for applying column restrictions to sub-matrices.
 
@@ -322,8 +322,8 @@ def getcol(self, i: int) -> Union[np.ndarray, sps.csr_matrix]:
     def sandwich(
         self,
         d: Union[np.ndarray, list],
-        rows: np.ndarray = None,
-        cols: np.ndarray = None,
+        rows: Optional[np.ndarray] = None,
+        cols: Optional[np.ndarray] = None,
     ) -> np.ndarray:
         """Perform a sandwich product: X.T @ diag(d) @ X."""
         if np.shape(d) != (self.shape[0],):
@@ -370,7 +370,10 @@ def _get_col_stds(self, weights: np.ndarray, col_means: np.ndarray) -> np.ndarra
         return col_stds
 
     def matvec(
-        self, v: np.ndarray, cols: np.ndarray = None, out: np.ndarray = None
+        self,
+        v: np.ndarray,
+        cols: Optional[np.ndarray] = None,
+        out: Optional[np.ndarray] = None,
     ) -> np.ndarray:
         """Perform self[:, cols] @ other[cols]."""
         assert not isinstance(v, sps.spmatrix)
@@ -393,6 +396,7 @@ def matvec(
         # as the target for storing the final output. This reduces the number
         # of output arrays allocated from 2 to 1.
         is_matrix_dense = [isinstance(m, DenseMatrix) for m in self.matrices]
+        dense_matrix_idx: Union[int, np.intp]
         if np.any(is_matrix_dense):
             dense_matrix_idx = np.argmax(is_matrix_dense)
             sub_cols = subset_cols[dense_matrix_idx]
@@ -411,14 +415,14 @@ def matvec(
                 continue
             in_vec = v[idx, ...]
             mat.matvec(in_vec, sub_cols, out=out)
-        return out
+        return out  # type: ignore
 
     def transpose_matvec(
         self,
         v: Union[np.ndarray, list],
-        rows: np.ndarray = None,
-        cols: np.ndarray = None,
-        out: np.ndarray = None,
+        rows: Optional[np.ndarray] = None,
+        cols: Optional[np.ndarray] = None,
+        out: Optional[np.ndarray] = None,
     ) -> np.ndarray:
         """
         Perform: self[rows, cols].T @ vec[rows].
diff --git a/src/tabmat/standardized_mat.py b/src/tabmat/standardized_mat.py
index 77f39766..6ca7e97d 100644
--- a/src/tabmat/standardized_mat.py
+++ b/src/tabmat/standardized_mat.py
@@ -36,7 +36,7 @@ def __init__(
         self,
         mat: MatrixBase,
         shift: Union[np.ndarray, list],
-        mult: Union[np.ndarray, list] = None,
+        mult: Optional[Union[np.ndarray, list]] = None,
     ):
         shift_arr = np.atleast_1d(np.squeeze(shift))
         expected_shape = (mat.shape[1],)
@@ -48,14 +48,15 @@ def __init__(
             but it has shape {np.asarray(shift).shape}"""
             )
 
-        mult_arr = mult
-        if mult_arr is not None:
-            mult_arr = np.atleast_1d(np.squeeze(mult_arr))
+        if mult is not None:
+            mult_arr = np.atleast_1d(np.squeeze(mult))
             if not mult_arr.shape == expected_shape:
                 raise ValueError(
                     f"""Expected mult to be able to conform to shape {expected_shape},
                 but it has shape {np.asarray(mult).shape}"""
                 )
+        else:
+            mult_arr = None
 
         self.shift = shift_arr
         self.mult = mult_arr
@@ -67,8 +68,8 @@ def __init__(
     def matvec(
         self,
         other_mat: Union[np.ndarray, list],
-        cols: np.ndarray = None,
-        out: np.ndarray = None,
+        cols: Optional[np.ndarray] = None,
+        out: Optional[np.ndarray] = None,
     ) -> np.ndarray:
         """
         Perform self[:, cols] @ other[cols].
@@ -119,7 +120,10 @@ def getcol(self, i: int):
         return StandardizedMatrix(col, [self.shift[i]], mult)
 
     def sandwich(
-        self, d: np.ndarray, rows: np.ndarray = None, cols: np.ndarray = None
+        self,
+        d: np.ndarray,
+        rows: Optional[np.ndarray] = None,
+        cols: Optional[np.ndarray] = None,
     ) -> np.ndarray:
         """Perform a sandwich product: X.T @ diag(d) @ X."""
         if not hasattr(d, "dtype"):
@@ -169,9 +173,9 @@ def unstandardize(self) -> MatrixBase:
     def transpose_matvec(
         self,
         other: Union[np.ndarray, list],
-        rows: np.ndarray = None,
-        cols: np.ndarray = None,
-        out: np.ndarray = None,
+        rows: Optional[np.ndarray] = None,
+        cols: Optional[np.ndarray] = None,
+        out: Optional[np.ndarray] = None,
     ) -> np.ndarray:
         """
         Perform: self[rows, cols].T @ vec[rows].