diff --git a/src/transformers/models/vjepa2/modeling_vjepa2.py b/src/transformers/models/vjepa2/modeling_vjepa2.py
index bde505b4ea54..eedc94b845a4 100644
--- a/src/transformers/models/vjepa2/modeling_vjepa2.py
+++ b/src/transformers/models/vjepa2/modeling_vjepa2.py
@@ -1125,7 +1125,7 @@ def forward(
         return encoder_output
 
     def get_vision_features(self, pixel_values_videos) -> torch.Tensor:
-        encoder_output = self.forward(pixel_values_videos)
+        encoder_output = self.forward(pixel_values_videos, skip_predictor=True)
         return encoder_output.last_hidden_state