switched ollama model, added script to add pdfs to the vector store, tuned RAG parameters

server.py

@@ -12,7 +12,7 @@ from local_rag import LocalRAG
 
 # Initialize RAG once at startup
 VECTORSTORE_PATH = "./vectorstore"
-OLLAMA_MODEL = "mistral:7b"
+OLLAMA_MODEL = "gpt-oss:20b"
 rag = LocalRAG(vectorstore_path=VECTORSTORE_PATH, ollama_model=OLLAMA_MODEL)
 
 app = FastAPI(title="Local RAG Chat", version="1.0.0")
@@ -32,6 +32,7 @@ class RetrievedChunk(BaseModel):
     content: str
     source: str
     page: int | None
+    score: float | None = None  # L2 distance from FAISS (lower = more similar)
 
 
 class ChatResponse(BaseModel):