diff --git a/pdf_agents/scientific_value.py b/pdf_agents/scientific_value.py
index fec2a70..d73da03 100644
--- a/pdf_agents/scientific_value.py
+++ b/pdf_agents/scientific_value.py
@@ -13,7 +13,6 @@
 from scipy.spatial import distance_matrix
 
 from .agents import PDFBaseAgent
-from .sklearn import PassiveKmeansAgent
 
 logger = getLogger("pdf_agents.scientific_value")
 
@@ -88,7 +87,7 @@ def __init__(
             if device is None
             else torch.device(device)
         )
-        self.bounds = torch.tensor(bounds, device=self.device).view(2, -1)
+        self.bounds = torch.tensor(bounds, device=self.device, dtype=torch.float).view(2, -1)
 
         self.num_restarts = num_restarts
         self.raw_samples = raw_samples
@@ -152,12 +151,22 @@ def ask(self, batch_size: int = 1):
                 candidate=candidate.detach().cpu().numpy(),
                 acquisition_value=acq.detach().cpu().numpy(),
                 latest_data=self.tell_cache[-1],
-                cache_len=len(self.independent_cache),
+                cache_len=(
+                    len(self.independent_cache)
+                    if isinstance(self.independent_cache, list)
+                    else self.independent_cache.shape[0]
+                ),
                 latest_value=value.squeeze()[-1],
                 ucb_beta=self.ucb_beta,
+                absolute_position_offset=self._motor_origins,
             )
             for candidate, acq in zip(candidates, acq_value)
         ]
         if not hasattr(self.independent_cache[0], "shape"):
             candidates = candidates.squeeze()
         return docs, torch.atleast_1d(candidates).detach().cpu().numpy().tolist()
+
+    def measurement_plan(self, relative_point):
+        """Send measurement plan absolute point from reltive position"""
+        absolute_point = relative_point + self._motor_origins
+        return super().measurement_plan(absolute_point)