lru cache & run on load

app.py

@@ -1,5 +1,6 @@
 from collections import defaultdict
 import spaces
+import functools
 import math
 import os
 import gradio as gr
@@ -463,6 +464,7 @@ except Exception as e:
 
 # --- Main Processing Function ---
 
+@functools.lru_cache(maxsize=8)
 @spaces.GPU
 def process_text(prompt: str) -> Tuple[
     Optional[matplotlib.figure.Figure], List[Tuple[str, str]], int, # BLT
@@ -620,6 +622,20 @@ with gr.Blocks(theme=Config.GRADIO_THEME) as iface:
             status_output
         ]
     )
+    iface.load(
+        fn=process_text,
+        inputs=prompt_input,
+        outputs=[
+            plot_output,
+            highlighted_output_bl,
+            bl_count_output,
+            highlighted_output_tk,
+            tk_count_output,
+            highlighted_output_llama,
+            llama_count_output,
+            status_output
+        ]
+    )
 
 # --- Launch the Gradio App ---
 if __name__ == "__main__":