diff --git a/tsimcne/tsimcne.py b/tsimcne/tsimcne.py
index f22d521..f4b3cdf 100644
--- a/tsimcne/tsimcne.py
+++ b/tsimcne/tsimcne.py
@@ -25,6 +25,7 @@ def __init__(
         n_epochs=100,
         batch_size=512,
         out_dim=2,
+        pretrain_out_dim=128,
         optimizer_name="sgd",
         lr_scheduler_name="cos_annealing",
         lr="auto_batch",
@@ -41,6 +42,7 @@ def __init__(
         self.projection_head = projection_head
         self.n_epochs = n_epochs
         self.out_dim = out_dim
+        self.pretrain_out_dim = pretrain_out_dim
         self.batch_size = batch_size
         self.optimizer_name = optimizer_name
         self.lr_scheduler_name = lr_scheduler_name
@@ -57,7 +59,7 @@ def _handle_parameters(self):
             self.model = make_model(
                 backbone=self.backbone,
                 projection_head=self.projection_head,
-                out_dim=self.out_dim,
+                out_dim=self.pretrain_out_dim,
             )
 
         if self.loss == "infonce":
@@ -201,6 +203,10 @@ class TSimCNE:
         of dimensions (so it could also be used to plot a dataset in
         3D, for example).
 
+
+    :param 128 pretrain_out_dim: The number of output dimensions
+        during pretraining (the first stage).
+
     :param "sgd" optimizer: The optimizer to use.  Currently only
         ``"sgd"`` is allowed.
 
@@ -255,6 +261,7 @@ def __init__(
         total_epochs=[1000, 50, 450],
         batch_size=512,
         out_dim=2,
+        pretrain_out_dim=128,
         optimizer="sgd",
         lr_scheduler="cos_annealing",
         lr="auto_batch",
@@ -272,6 +279,7 @@ def __init__(
         self.projection_head = projection_head
         self.data_transform = data_transform
         self.out_dim = out_dim
+        self.pretrain_out_dim = pretrain_out_dim
         self.batch_size = batch_size
         self.optimizer = optimizer
         self.lr_scheduler = lr_scheduler
@@ -440,7 +448,7 @@ def fit(self, X: torch.utils.data.Dataset | str):
                     metric=self.metric,
                     backbone=self.backbone,
                     projection_head=self.projection_head,
-                    out_dim=128,
+                    out_dim=self.pretrain_out_dim,
                     **train_kwargs,
                 )