Avoid batch recommendation with marginal surrogate models

examples/Custom_Surrogates/surrogate_params.py

@@ -97,7 +97,7 @@
 print(surrogate_model.to_json(), end="\n" * 3)
 
 # Let's do a first round of recommendation
-recommendation = campaign.recommend(batch_size=2)
+recommendation = campaign.recommend(batch_size=1)
 
 print("Recommendation from campaign:")
 print(recommendation)
@@ -112,8 +112,8 @@
 
 print("Here you will see some model outputs as we set verbose to True")
 
-# Do another round of recommendations
-recommendation = campaign.recommend(batch_size=2)
+# Do another round of recommendation
+recommendation = campaign.recommend(batch_size=1)
 
 
 # Print second round of recommendations

tests/test_iterations.py

@@ -37,7 +37,7 @@
 )
 from baybe.recommenders.pure.nonpredictive.base import NonPredictiveRecommender
 from baybe.searchspace import SearchSpaceType
-from baybe.surrogates.base import Surrogate
+from baybe.surrogates.base import IndependentGaussianSurrogate, Surrogate
 from baybe.surrogates.custom import CustomONNXSurrogate
 from baybe.surrogates.gaussian_process.presets import (
     DefaultKernelFactory,
@@ -249,7 +249,9 @@ def test_kernel_factories(campaign, n_iterations, batch_size):
     valid_surrogate_models,
     ids=[c.__class__ for c in valid_surrogate_models],
 )
-def test_surrogate_models(campaign, n_iterations, batch_size):
+def test_surrogate_models(campaign, n_iterations, batch_size, surrogate_model):
+    if batch_size > 1 and isinstance(surrogate_model, IndependentGaussianSurrogate):
+        pytest.skip("Batch recommendation is not supported.")
     run_iterations(campaign, n_iterations, batch_size)