Merge remote-tracking branch 'origin/main'

# Conflicts:
#	mm_agents/gpt_4v_agent.py

mm_agents/gpt_4v_agent.py

@@ -30,6 +30,8 @@ from mm_agents.prompts import SYS_PROMPT_IN_SCREENSHOT_OUT_CODE, SYS_PROMPT_IN_S
     SYS_PROMPT_IN_SOM_A11Y_OUT_TAG, \
     SYS_PROMPT_SEEACT, ACTION_DESCRIPTION_PROMPT_SEEACT, ACTION_GROUNDING_PROMPT_SEEACT
 
+import logging
+
 logger = logging.getLogger("desktopenv.agent")
 
 
@@ -165,9 +167,6 @@ def parse_code_from_som_string(input_string, masks):
                 logger.error("Predicting the tag with index {} failed.".format(match))
                 return ""
 
-        return text
-
-    input_string = replace_tags_with_mappings(input_string, mappings)
     actions = parse_code_from_string(input_string)
     return actions
 
@@ -479,11 +478,14 @@ class GPT4v_Agent:
         with open("messages.json", "w") as f:
             f.write(json.dumps(messages, indent=4))
 
-        response = self.call_llm({
-            "model": self.model,
-            "messages": messages,
-            "max_tokens": self.max_tokens
-        })
+        try:
+            response = self.call_llm({
+                "model": self.model,
+                "messages": messages,
+                "max_tokens": self.max_tokens
+            })
+        except:
+            response = ""
 
         logger.debug("RESPONSE: %s", response)