update

models/vsa_model.py

@@ -372,6 +372,7 @@ class VisionSearchAssistant:
         
         del self.grounder
         torch.cuda.empty_cache()
+        torch.cuda.synchronize()
         
         # Visual Captioning
         self.vlm = VLM(
@@ -415,6 +416,7 @@ class VisionSearchAssistant:
 
         del self.vlm
         torch.cuda.empty_cache()
+        torch.cuda.synchronize()
 
         # Web Searching
         contexts = self.searcher(queries)
@@ -442,4 +444,5 @@ class VisionSearchAssistant:
         yield answer, 'answer'
 
         del self.vlm
-        torch.cuda.empty_cache()
+        torch.cuda.empty_cache()
+        torch.cuda.synchronize()