Update app.py

app.py

@@ -110,7 +110,7 @@ with gr.Blocks(
         
         This Space showcases a multimodal captioning model built from:
         
-        - **CLIP ViT-L/14 encoder (last 4 layers fine-tuned)**
+        - **CLIP ViT-B/16 encoder (last 4 layers fine-tuned)**
         - **T5-Base decoder (fully fine-tuned)**
         
         This is a mid-tier model from within our experiments, but due to compute limitations on the demo server, this configuration provides a good balance