Update src/generation/llm.py

src/generation/llm.py

@@ -100,7 +100,7 @@ class LLM:
             # Decode the generated tokens
             response = self.tokenizer.decode(outputs[0], skip_special_tokens=True)
             response = response.strip()
-            keyword = "assistant"
+            keyword = "Truy vấn:"
             index = response.find(keyword)
             output = response[index + len(keyword):].strip() if index != -1 else response
             print("Response generated successfully!")