Update app.py

app.py

@@ -17,14 +17,6 @@ from gradio_helper import make_demo  # UI logic import
 from llm_config import SUPPORTED_LLM_MODELS
 
 # Model configuration setup
-model_name = model_configuration["model_id"]
-start_message = model_configuration["start_message"]
-history_template = model_configuration.get("history_template")
-has_chat_template = model_configuration.get("has_chat_template", history_template is None)
-current_message_template = model_configuration.get("current_message_template")
-stop_tokens = model_configuration.get("stop_tokens")
-tokenizer_kwargs = model_configuration.get("tokenizer_kwargs", {})
-
 max_new_tokens = 256
 model_language_value = "English"
 model_id_value = 'qwen2.5-0.5b-instruct'
@@ -37,6 +29,15 @@ pt_model_name = model_id_value.split("-")[0]
 int4_model_dir = Path(model_id_value) / "INT4_compressed_weights"
 int4_weights = int4_model_dir / "openvino_model.bin"
 
+model_configuration = SUPPORTED_LLM_MODELS[model_language_value][model_id_value]
+model_name = model_configuration["model_id"]
+start_message = model_configuration["start_message"]
+history_template = model_configuration.get("history_template")
+has_chat_template = model_configuration.get("has_chat_template", history_template is None)
+current_message_template = model_configuration.get("current_message_template")
+stop_tokens = model_configuration.get("stop_tokens")
+tokenizer_kwargs = model_configuration.get("tokenizer_kwargs", {})
+
 # Model loading
 core = ov.Core()
 ov_config = {