Drop attrs typing workaround

baybe/parameters/substance.py

@@ -6,7 +6,7 @@
 
 import pandas as pd
 from attrs import define, field
-from attrs.validators import and_, deep_mapping, instance_of, min_len
+from attrs.validators import deep_mapping, instance_of, min_len
 from typing_extensions import override
 
 from baybe.parameters.base import _DiscreteLabelLikeParameter
@@ -44,9 +44,7 @@ class SubstanceParameter(_DiscreteLabelLikeParameter):
         converter=lambda x: dict(sorted(x.items())),
         validator=deep_mapping(
             mapping_validator=min_len(2),
-            # FIXME[typing]: https://github.com/python-attrs/attrs/issues/1206
-            key_validator=and_(instance_of(str), min_len(1)),
-            value_validator=lambda *x: None,
+            key_validator=(instance_of(str), min_len(1)),
         ),
     )
     """A mapping that provides the SMILES strings for all available parameter values."""

baybe/transformations/composite.py

@@ -7,7 +7,7 @@
 from typing import TYPE_CHECKING, Any
 
 from attrs import define, field
-from attrs.validators import and_, deep_iterable, instance_of, max_len, min_len
+from attrs.validators import deep_iterable, instance_of, max_len, min_len
 from typing_extensions import override
 
 from baybe.transformations.base import Transformation
@@ -69,7 +69,7 @@ class AdditiveTransformation(Transformation):
     transformations: tuple[Transformation, Transformation] = field(
         converter=to_tuple,
         validator=deep_iterable(
-            iterable_validator=and_(min_len(2), max_len(2)),
+            iterable_validator=(min_len(2), max_len(2)),
             member_validator=instance_of(Transformation),
         ),
     )
@@ -94,7 +94,7 @@ class MultiplicativeTransformation(Transformation):
     transformations: tuple[Transformation, Transformation] = field(
         converter=to_tuple,
         validator=deep_iterable(
-            iterable_validator=and_(min_len(2), max_len(2)),
+            iterable_validator=(min_len(2), max_len(2)),
             member_validator=instance_of(Transformation),
         ),
     )

baybe/utils/metadata.py

@@ -31,8 +31,6 @@ class Metadata(SerialMixin):
         validator=deep_mapping(
             mapping_validator=instance_of(dict),
             key_validator=instance_of(str),
-            # FIXME: https://github.com/python-attrs/attrs/issues/1246
-            value_validator=lambda *x: None,
         ),
         kw_only=True,
     )

benchmarks/definition/regression.py

@@ -1,7 +1,7 @@
 """Regression benchmark configuration."""
 
 from attrs import define, field
-from attrs.validators import and_, deep_iterable, ge, instance_of, le
+from attrs.validators import deep_iterable, ge, instance_of, le
 
 from baybe.utils.validation import finite_float
 from benchmarks.definition.base import Benchmark, BenchmarkSettings
@@ -17,7 +17,7 @@ class RegressionBenchmarkSettings(BenchmarkSettings):
     max_n_train_points: int = field(validator=instance_of(int))
     """Maximum number of training points to consider."""
 
-    noise_std: float = field(converter=float, validator=and_(finite_float, ge(0.0)))
+    noise_std: float = field(converter=float, validator=[finite_float, ge(0.0)])
     """Standard deviation of noise to add to the data."""
 
 
@@ -27,7 +27,7 @@ class TransferLearningRegressionBenchmarkSettings(RegressionBenchmarkSettings):
 
     source_fractions: tuple[float, ...] = field(
         validator=deep_iterable(
-            member_validator=and_(ge(0.0), le(1.0)),
+            member_validator=(ge(0.0), le(1.0)),
             iterable_validator=instance_of(tuple),
         )
     )

pyproject.toml

@@ -29,7 +29,7 @@ keywords = [
 ]
 dynamic = ['version']
 dependencies = [
-    "attrs>=24.1.0",
+    "attrs>=25.4.0",
     "botorch>=0.13.0,<1",
     "cattrs>=25.2.0",
     "exceptiongroup",