Fix: Use MiniLM-L6-v2 to match Qdrant vector dimension (384)

app.py

@@ -11,6 +11,7 @@ from src.data_pipeline import get_embedding
 COLLECTION_NAME = "dashVector_v1"
 VECTOR_SIZE = 384 # MiniLM-L6-v2
 NUM_CLUSTERS = 32
+EMBEDDING_MODEL = "sentence-transformers/all-MiniLM-L6-v2"
 
 # --- Initialize Backend ---
 # We initialize once at startup
@@ -251,7 +252,7 @@ def run_benchmark(query):
         
         # Generate Embedding
         print("DEBUG: Generating embedding...")
-        query_vec = get_embedding(query)
+        query_vec = get_embedding(query, model_name=EMBEDDING_MODEL)
         print("DEBUG: Embedding generated.")
 
         # Router Prediction