diff --git a/pymc3/model.py b/pymc3/model.py
index d68705148c1..95140d30af4 100644
--- a/pymc3/model.py
+++ b/pymc3/model.py
@@ -67,10 +67,10 @@
 from pymc3.util import (
     UNSET,
     WithMemoization,
+    get_transformed_name,
     get_var_name,
     treedict,
     treelist,
-    get_transformed_name,
 )
 from pymc3.vartypes import continuous_types, discrete_types, typefilter
 
@@ -943,7 +943,8 @@ def test_point(self) -> Dict[str, np.ndarray]:
     @property
     def initial_point(self) -> Dict[str, np.ndarray]:
         """Maps free variable names to transformed, numeric initial values."""
-        return self.recompute_initial_point()
+        seed = self.rng_seeder.randint(2 ** 30, dtype=np.int64)
+        return self.recompute_initial_point(seed)
 
     def recompute_initial_point(self, seed) -> Dict[str, np.ndarray]:
         """Recomputes the initial point of the model.
@@ -959,9 +960,8 @@ def recompute_initial_point(self, seed) -> Dict[str, np.ndarray]:
     def _make_initial_point_expression(
         self,
         *,
-        rng: Optional[Union[int, np.random.SeedSequence]] = None,
         jitter_rvs: Set[TensorVariable] = None,
-        default_strategy: str = "moment",
+        default_strategy: str = "prior",
         return_transformed: bool = False,
     ) -> List[TensorVariable]:
         """Recomputes numeric initial values for all free model variables.
@@ -999,33 +999,23 @@ def _make_initial_point_expression(
             elif strategy == "prior":
                 value = variable
             else:
-                value = strategy
+                value = at.as_tensor_variable(strategy)
 
             transform = getattr(self.rvs_to_values[variable].tag, "transform", None)
 
-            print("-----")
-
             if transform is not None:
                 value = transform.forward(variable, value)
 
-            print("transform", transform)
-            print("value", value)
-
             if variable in jitter_rvs:
                 jitter = at.random.uniform(at.zeros_like(variable) - 1, at.zeros_like(variable) + 1)
                 jitter.name = "jitter"
                 value = value + jitter
 
-            print("after_jitter", value)
-
             initial_values_transformed.append(value)
 
             if transform is not None:
                 value = transform.backward(variable, value)
 
-            print("final", value)
-            print()
-
             initial_values.append(value)
 
         # Copy the outputs in a way that we still know what corresponds to the free_RVS
@@ -1065,9 +1055,9 @@ def make_initial_point_fn(
         self,
         *,
         jitter_rvs: Set[TensorVariable] = None,
-        default_strategy: str = "moment",
+        default_strategy: str = "prior",
         return_transformed: bool,
-    ) -> Callable[Union[int, np.random.SeedSequence], Dict[str, np.ndarray]]:
+    ) -> Callable[[Union[int, np.random.SeedSequence]], Dict[str, np.ndarray]]:
         """Recomputes numeric initial values for all free model variables.
 
         Parameters
diff --git a/pymc3/tests/test_initvals.py b/pymc3/tests/test_initvals.py
index 9ff8edd3291..41c0fe7b36c 100644
--- a/pymc3/tests/test_initvals.py
+++ b/pymc3/tests/test_initvals.py
@@ -124,7 +124,6 @@ def test_adds_jitter(self):
         assert iv["A"] == 0
         # Moment of the HalfFlat is 1, but HalfFlat is log-transformed by default
         # so the transformed initial value with jitter will be
-        print(iv)
         b_transformed = iv["B_log__"]
         b_untransformed = transform_back(B, b_transformed)
         assert b_transformed != 0
diff --git a/pymc3/tests/test_model.py b/pymc3/tests/test_model.py
index 0537ada9537..5e30dccbecd 100644
--- a/pymc3/tests/test_model.py
+++ b/pymc3/tests/test_model.py
@@ -517,7 +517,7 @@ def test_initial_point():
     assert a in model.initial_values
     assert x in model.initial_values
     assert model.initial_values[b] == b_initval
-    assert model.recompute_initial_point()["b_interval__"] == b_initval_trans
+    assert model.recompute_initial_point(0)["b_interval__"] == b_initval_trans
     assert model.initial_values[y] == y_initval