diff --git a/metalearners/_typing.py b/metalearners/_typing.py
index 03f6192..a7f39d4 100644
--- a/metalearners/_typing.py
+++ b/metalearners/_typing.py
@@ -1,7 +1,7 @@
 # # Copyright (c) QuantCo 2024-2024
 # # SPDX-License-Identifier: BSD-3-Clause
 
-from collections.abc import Collection
+from collections.abc import Collection, Mapping
 from typing import Literal, Protocol, Union
 
 import numpy as np
@@ -16,7 +16,7 @@
 # https://mypy.readthedocs.io/en/stable/literal_types.html#limitations
 OosMethod = Literal["overall", "median", "mean"]
 
-Params = dict[str, int | float | str]
+Params = Mapping[str, int | float | str]
 Features = Collection[str] | Collection[int]
 
 # ruff is not happy about the usage of Union.
diff --git a/tests/test_learner.py b/tests/test_learner.py
index df7c6da..e76018e 100644
--- a/tests/test_learner.py
+++ b/tests/test_learner.py
@@ -588,13 +588,13 @@ def test_conditional_average_outcomes_smoke(
     factory = metalearner_factory(metalearner_prefix)
     learner = factory(
         nuisance_model_factory=_tree_base_learner(is_classification),
-        nuisance_model_params={"n_estimators": 1},  # type: ignore
+        nuisance_model_params={"n_estimators": 1},
         is_classification=is_classification,
         n_variants=len(np.unique(df[treatment_column])),
         treatment_model_factory=LGBMRegressor,
-        treatment_model_params={"n_estimators": 1},  # type: ignore
+        treatment_model_params={"n_estimators": 1},
         propensity_model_factory=LGBMClassifier,
-        propensity_model_params={"n_estimators": 1},  # type: ignore
+        propensity_model_params={"n_estimators": 1},
         n_folds=2,
     )
     learner.fit(df[feature_columns], df[outcome_column], df[treatment_column])
@@ -626,7 +626,7 @@ def test_conditional_average_outcomes_smoke_multi_class(
     y = rng.integers(0, n_classes, size=sample_size)
     learner = factory(
         nuisance_model_factory=_tree_base_learner(True),
-        nuisance_model_params={"n_estimators": 1},  # type: ignore
+        nuisance_model_params={"n_estimators": 1},
         n_variants=n_variants,
         is_classification=True,
         n_folds=2,
@@ -665,13 +665,13 @@ def test_predict_smoke(
         y = rng.standard_normal(sample_size)
     learner = factory(
         nuisance_model_factory=_tree_base_learner(is_classification),
-        nuisance_model_params={"n_estimators": 1},  # type: ignore
+        nuisance_model_params={"n_estimators": 1},
         n_variants=n_variants,
         is_classification=is_classification,
         treatment_model_factory=LGBMRegressor,
-        treatment_model_params={"n_estimators": 1},  # type: ignore
+        treatment_model_params={"n_estimators": 1},
         propensity_model_factory=LGBMClassifier,
-        propensity_model_params={"n_estimators": 1},  # type: ignore
+        propensity_model_params={"n_estimators": 1},
         n_folds=2,
     )
     learner.fit(X, y, w)
diff --git a/tests/test_metalearner.py b/tests/test_metalearner.py
index 3fa4015..1e41423 100644
--- a/tests/test_metalearner.py
+++ b/tests/test_metalearner.py
@@ -737,7 +737,7 @@ def test_feature_importances_smoke(
         nuisance_model_factory=LinearRegression,
         treatment_model_factory=LGBMRegressor,
         propensity_model_factory=LogisticRegression,
-        treatment_model_params={"n_estimators": 1},  # type: ignore
+        treatment_model_params={"n_estimators": 1},
     )
 
     ml.fit(X=X, y=y, w=w)
@@ -895,7 +895,7 @@ def test_shap_values_smoke(
         nuisance_model_factory=LinearRegression,
         treatment_model_factory=LGBMRegressor,
         propensity_model_factory=LogisticRegression,
-        treatment_model_params={"n_estimators": 1},  # type: ignore
+        treatment_model_params={"n_estimators": 1},
     )
 
     ml.fit(X=X, y=y, w=w)