diff --git a/trulens/nn/models/__init__.py b/trulens/nn/models/__init__.py
index ae23f4873..08c03da07 100644
--- a/trulens/nn/models/__init__.py
+++ b/trulens/nn/models/__init__.py
@@ -74,6 +74,7 @@ def get_model_wrapper(
     default_feed_dict=None,
     session=None,
     backend=None,
+    force_eval=True,
     **kwargs
 ):
     """
@@ -136,6 +137,10 @@ def get_model_wrapper(
         backend:
             _Optional, for forcing a specific backend._ String values recognized
             are pytorch, tensorflow, keras, or tf.keras.
+        
+        force_eval:
+            _Optional, True will force a model.eval() call for PyTorch models. False
+            will retain current model state
 
     Returns: ModelWrapper
     """
@@ -189,7 +194,10 @@ def get_model_wrapper(
     elif B.backend == Backend.PYTORCH:
         from trulens.nn.models.pytorch import PytorchModelWrapper
         return PytorchModelWrapper(
-            model, logit_layer=logit_layer, device=device
+            model,
+            logit_layer=logit_layer,
+            device=device,
+            force_eval=force_eval
         )
     elif B.backend == Backend.TENSORFLOW:
         import tensorflow as tf
diff --git a/trulens/nn/models/pytorch.py b/trulens/nn/models/pytorch.py
index cad6ddeec..339ae5bcb 100644
--- a/trulens/nn/models/pytorch.py
+++ b/trulens/nn/models/pytorch.py
@@ -33,7 +33,15 @@ class PytorchModelWrapper(ModelWrapper):
     of Pytorch nn.Module objects.
     """
 
-    def __init__(self, model, *, logit_layer=None, device=None, **kwargs):
+    def __init__(
+        self,
+        model,
+        *,
+        logit_layer=None,
+        device=None,
+        force_eval=True,
+        **kwargs
+    ):
         """
         __init__ Constructor
 
@@ -46,6 +54,9 @@ def __init__(self, model, *, logit_layer=None, device=None, **kwargs):
             layer named 'logits' is the logit layer.
         device : string, optional
             device on which to run model, by default None
+        force_eval : bool, optional
+            If True, will call model.eval() to ensure determinism. Otherwise, keeps current model state, by default True
+            
         """
 
         if 'input_shape' in kwargs:
@@ -61,8 +72,9 @@ def __init__(self, model, *, logit_layer=None, device=None, **kwargs):
 
         super().__init__(model, **kwargs)
         # sets self._model, issues cross-backend messages
-
-        model.eval()
+        self.force_eval = force_eval
+        if self.force_eval:
+            model.eval()
 
         if device is None:
             device = pytorch.get_default_device()
@@ -337,7 +349,8 @@ def hookfn(self, inpt, outpt):
         with memory_suggestions(device=self.device):
             # Run the network.
             try:
-                self._model.eval()  # needed for determinism sometimes
+                if self.force_eval:
+                    self._model.eval()  # needed for determinism sometimes
                 output = model_inputs.call_on(self._model)
 
                 if isinstance(output, tuple):