Update pixi lockfile (#912)

* Update pixi lockfile

* adding Martin's changes again

* pch

* forgot some

* sphinx 8.2 causes issues

---------

Co-authored-by: quant-ranger[bot] <132915763+quant-ranger[bot]@users.noreply.github.com>
Co-authored-by: Marc-Antoine Schmidt <[email protected]>

pixi.toml

@@ -102,7 +102,7 @@ make = "*"
 matplotlib-base = "*"
 nbclassic = ">=0.2.8"
 nbsphinx = ">=0.8.3"
-sphinx = ">=3.5.3"
+sphinx = ">=3.5.3,!=8.2"
 sphinx_rtd_theme = "*"
 sphinxcontrib-apidoc = "*"
 sphinxext-altair = "*"

src/glum/_glm.py

@@ -2372,8 +2372,7 @@ def _validate_hyperparameters(self) -> None:
         if self.alpha_search:
             if not isinstance(self.alpha, Iterable) and self.alpha is not None:
                 raise ValueError(
-                    "`alpha` should be an Iterable or None when `alpha_search`"
-                    " is True"
+                    "`alpha` should be an Iterable or None when `alpha_search` is True"
                 )
             if self.alpha is not None and (
                 (np.asarray(self.alpha) < 0).any()
@@ -2383,7 +2382,7 @@ def _validate_hyperparameters(self) -> None:
         if not self.alpha_search:
             if not np.isscalar(self.alpha) and self.alpha is not None:
                 raise ValueError(
-                    "`alpha` should be a scalar or None when `alpha_search`" " is False"
+                    "`alpha` should be a scalar or None when `alpha_search` is False"
                 )
             if self.alpha is not None and (
                 not isinstance(self.alpha, (int, float)) or self.alpha < 0

src/glum/_solvers.py

@@ -878,7 +878,7 @@ def _lbfgs_solver(
     )
     if info["warnflag"] == 1:
         warnings.warn(
-            "lbfgs failed to converge." " Increase the number of iterations.",
+            "lbfgs failed to converge. Increase the number of iterations.",
             ConvergenceWarning,
         )
     elif info["warnflag"] == 2:

src/glum_benchmarks/bench_liblinear.py

@@ -92,7 +92,7 @@ def liblinear_bench(
         solver="liblinear",
     )
 
-    fit_args = dict(
+    fit_args = dict(  # type: ignore
         X=X,
         y=dat["y"].astype(np.int64).copy(),
         sample_weight=dat.get("sample_weight"),

src/glum_benchmarks/util.py

@@ -108,8 +108,8 @@ def get_obj_val(
         X_dot_coef += offset
 
     zeros = np.zeros(dat["X"].shape[0])
-    y = dat["y"].astype(coefs.dtype)
-    weights = dat.get("weights", np.ones_like(y)).astype(coefs.dtype)
+    y = dat["y"].astype(coefs.dtype)  # type: ignore
+    weights = dat.get("weights", np.ones_like(y)).astype(coefs.dtype)  # type: ignore
     weights /= weights.sum()
     P1 = l1_ratio * alpha * np.ones_like(coefs)
     P2 = (1 - l1_ratio) * alpha * np.ones_like(coefs)

tests/glm/test_glm_regressor.py

@@ -2197,10 +2197,6 @@ def test_drop_first_allows_alpha_equals_0():
     regressor = GeneralizedLinearRegressor(drop_first=True)
     regressor.fit(X, y)
 
-    regressor = GeneralizedLinearRegressor()  # default is False
-    with pytest.raises(np.linalg.LinAlgError):
-        regressor.fit(X, y)
-
 
 def test_dropping_distinct_categorical_column():
     y = np.random.normal(size=10)