Minor memory tweaks

model_setup.py

@@ -2,10 +2,11 @@
 """loading the models to be used by the Mulltimodal RAG system."""
 
 import torch
+import gc
 
 from sentence_transformers import SentenceTransformer
 from transformers import AutoProcessor, Gemma3ForConditionalGeneration, BitsAndBytesConfig
-
+from utils import clear_gpu_cache
 device = "cuda" if torch.cuda.is_available() else "cpu"
 
 # Embedding model
@@ -33,3 +34,7 @@ model.eval()
 
 # Processor
 processor = AutoProcessor.from_pretrained(model_name, use_fast=True)
+
+# Free memory
+clear_gpu_cache()
+gc.collect()