fixed streaming for ctransformers

app_modules/llm_loader.py

@@ -202,7 +202,10 @@ class LLMLoader:
                     "repetition_penalty": 1.1,
                 }
                 self.llm = CTransformers(
-                    model=MODEL_PATH, model_type="llama", config=config
+                    model=MODEL_PATH,
+                    model_type="llama",
+                    config=config,
+                    callbacks=callbacks,
                 )
             elif self.llm_model_type == "hftgi":
                 HFTGI_SERVER_URL = os.environ.get("HFTGI_SERVER_URL")