Update app.py

app.py

@@ -128,7 +128,10 @@ def load_models_and_documents():
             n_gpu_layers=32
         )
 
-        dense_model = OptimumEncoder(cache_dir=os.getenv('HF_HOME'))
+        dense_model = OptimumEncoder(
+            device='cuda',
+            cache_dir=os.getenv('HF_HOME')
+        )
         
         sparse_model = Bm42(
             'Qdrant/all_miniLM_L6_v2_with_attentions',