Update app.py

app.py

@@ -125,7 +125,8 @@ async def lifespan(app: FastAPI):
         ml_models["reranker_compressor"] = CrossEncoderReranker(model=cross_encoder_model, top_n=9)
         
         ml_models["llm"] = ChatGoogleGenerativeAI(
-                               model="gemini-2.5-flash", #using flash 2.5 as its give better result but slower than lower flash
+                               # model="gemini-2.5-flash", #using flash 2.5 as its give better result but slower than lower flash
+                               model="gemini-1.5-flash", 
                                api_key=GOOGLE_API_KEY,
                            )
         
@@ -148,7 +149,7 @@ Step 1 – **Initial Draft**:
 - Summarize relevant parts of the context without losing meaning.
 - Avoid boilerplate phrases like “the document states” or “according to the context.”
 - If the answer is not in the context for some subqueries, respond exactly with: " I do not know the answer of "subquery",Please ask query related to the Document only." for that subquery.
-- If the query is not in english language then respond excatly with: "I do not know this language , ask in English Only".
+- Make sure that the You answer  the query in the same language in which the query is asked.
 
 Step 2 – **Critique & Revise**:
 - Review the initial answers for any missing or underused context.