update

app/models/llm.py

@@ -17,7 +17,8 @@ def load_model():
     _model = Llama(
         model_path=model_path,
         n_ctx=4096,
-        n_threads=4,
+        n_threads=2,        # Match this EXACTLY to the 2 vCPUs provided by free Spaces
+        n_threads_batch=2,  
         verbose=False,
     )