Update app.py

app.py

@@ -34,15 +34,6 @@ tokenizer = AutoTokenizer.from_pretrained(model_name)
 model = AutoModelForCausalLM.from_pretrained(model_name)
 tokenizer.pad_token = tokenizer.eos_token
 
-# Connect to Qdrant + embedding
-embedding = HuggingFaceEmbeddings(model_name="Omartificial-Intelligence-Space/GATE-AraBert-v1")
-qdrant_client = QdrantClient(url=QDRANT_URL, api_key=QDRANT_API_KEY)
-
-vector_store = Qdrant(
-    client=qdrant_client,
-    collection_name=COLLECTION_NAME,
-    embeddings=embedding
-)
 
 # Generation settings
 generation_config = GenerationConfig(
@@ -65,6 +56,16 @@ llm_pipeline = pipeline(
 
 llm = HuggingFacePipeline(pipeline=llm_pipeline)
 
+# Connect to Qdrant + embedding
+embedding = HuggingFaceEmbeddings(model_name="Omartificial-Intelligence-Space/GATE-AraBert-v1")
+qdrant_client = QdrantClient(url=QDRANT_URL, api_key=QDRANT_API_KEY)
+
+vector_store = Qdrant(
+    client=qdrant_client,
+    collection_name=COLLECTION_NAME,
+    embeddings=embedding
+)
+
 retriever = vector_store.as_retriever(search_kwargs={"k": 3})
 
 # Set up RAG QA chain