diff --git a/changelog_entry.yaml b/changelog_entry.yaml
index e69de29..8558889 100644
--- a/changelog_entry.yaml
+++ b/changelog_entry.yaml
@@ -0,0 +1,4 @@
+- bump: patch
+  changes:
+    fixed:
+    - Final weights are now consistent with the training log.
diff --git a/src/microcalibrate/reweight.py b/src/microcalibrate/reweight.py
index df40088..3145a70 100644
--- a/src/microcalibrate/reweight.py
+++ b/src/microcalibrate/reweight.py
@@ -20,7 +20,7 @@ def reweight(
     estimate_function: Callable[[Tensor], Tensor],
     targets_array: np.ndarray,
     target_names: np.ndarray,
-    dropout_rate: Optional[float] = 0.1,
+    dropout_rate: Optional[float] = 0.05,
     epochs: Optional[int] = 2_000,
     noise_level: Optional[float] = 10.0,
     learning_rate: Optional[float] = 1e-3,
@@ -96,22 +96,15 @@ def dropout_weights(weights: torch.Tensor, p: float) -> torch.Tensor:
     loss_over_epochs = []
     estimates_over_epochs = []
     pct_close_over_epochs = []
+    max_epochs = epochs - 1 if epochs > 0 else 0
     epochs = []
 
     for i in iterator:
         optimizer.zero_grad()
-        running_loss = None
-        for j in range(2):
-            weights_ = dropout_weights(weights, dropout_rate)
-            estimate = estimate_function(torch.exp(weights_))
-            l = loss(estimate, targets)
-            close = pct_close(estimate, targets)
-            if running_loss is None:
-                running_loss = l
-            else:
-                running_loss += l
-
-        l = running_loss / 2
+        weights_ = dropout_weights(weights, dropout_rate)
+        estimate = estimate_function(torch.exp(weights_))
+        l = loss(estimate, targets)
+        close = pct_close(estimate, targets)
 
         if i % progress_update_interval == 0:
             iterator.set_postfix(
@@ -139,8 +132,9 @@ def dropout_weights(weights: torch.Tensor, p: float) -> torch.Tensor:
                     f"({'improving' if loss_change > 0 else 'worsening'})"
                 )
 
-        l.backward()
-        optimizer.step()
+        if i != max_epochs - 1:
+            l.backward()
+            optimizer.step()
 
     tracker_dict = {
         "epochs": epochs,
@@ -160,7 +154,7 @@ def dropout_weights(weights: torch.Tensor, p: float) -> torch.Tensor:
 
     logger.info(f"Reweighting completed. Final sample size: {len(weights)}")
 
-    final_weights = torch.exp(weights).detach().cpu().numpy()
+    final_weights = torch.exp(weights_).detach().cpu().numpy()
 
     return (
         final_weights,