cr

handler.py

@@ -63,4 +63,9 @@ class EndpointHandler():
             indices.append(batch.nonzero(as_tuple=True)[0].tolist())
             vecs.append(batch[indices[-1]].tolist())
 
+        # release all the GPU memory cache that can be freed
+        if torch.cuda.is_available():
+            torch.cuda.empty_cache()
+
+
         return [indices, vecs]