Implement heuristic cutting on the accessibility tree to get the important nodes; Finish accessibility tree text agent

mm_agents/gemini_pro_vision_agent.py

@@ -1,3 +1,4 @@
+import time
 from typing import Dict, List
 
 import PIL.Image
@@ -66,13 +67,20 @@ class GeminiProV_Agent:
 
         print("Trajectory:", traj_to_show)
 
-        response = self.model.generate_content(
-            message_for_gemini,
-            generation_config={
-                "max_output_tokens": self.max_tokens,
-                "temperature": self.temperature
-            }
-        )
+        while True:
+            try:
+                response = self.model.generate_content(
+                    message_for_gemini,
+                    generation_config={
+                        "max_output_tokens": self.max_tokens,
+                        "temperature": self.temperature
+                    }
+                )
+                break
+            except:
+                print("Failed to generate response, retrying...")
+                time.sleep(5)
+                pass
 
         try:
             response_text = response.text