lzy/sci-gui-agent-benchmark
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Dev/uitars (#132)

Browse Source 
* init uitars

* change agent class name

* FIX: return bug in agent predict
This commit is contained in:
Shihao Liang
2025-02-14 11:17:37 +08:00
committed by GitHub
parent 75601efc6a
commit 339a13e1d5
1 changed files with 11 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
mm_agents/uitars_agent.py
View File

@@ -688,7 +688,8 @@ class UITARSAgent:
try_times -= 1 try_times -= 1
if prediction is None: if prediction is None:
return "client error", ["DONE"], [] return "client error", ["DONE"]
self.history_responses.append(prediction) self.history_responses.append(prediction)
self.thoughts.append(prediction) self.thoughts.append(prediction)
@@ -702,7 +703,7 @@ class UITARSAgent:
) )
except Exception as e: except Exception as e:
print(f"Parsing action error: {prediction}, with error:\n{e}") print(f"Parsing action error: {prediction}, with error:\n{e}")
return f"Parsing action error: {prediction}, with error:\n{e}", ["DONE"], [] return f"Parsing action error: {prediction}, with error:\n{e}", ["DONE"]
actions = [] actions = []
for parsed_response in parsed_responses: for parsed_response in parsed_responses:
@@ -710,19 +711,21 @@ class UITARSAgent:
if parsed_response["action_type"] == FINISH_WORD: if parsed_response["action_type"] == FINISH_WORD:
self.actions.append(actions) self.actions.append(actions)
return prediction, ["DONE"], parsed_responses
return prediction, ["DONE"]
elif parsed_response["action_type"] == WAIT_WORD: elif parsed_response["action_type"] == WAIT_WORD:
self.actions.append(actions) self.actions.append(actions)
return prediction, ["WAIT"], parsed_responses return prediction, ["WAIT"]
elif parsed_response["action_type"] == ENV_FAIL_WORD: elif parsed_response["action_type"] == ENV_FAIL_WORD:
self.actions.append(actions) self.actions.append(actions)
return prediction, ["FAIL"], parsed_responses return prediction, ["FAIL"]
elif parsed_response["action_type"] == CALL_USER: elif parsed_response["action_type"] == CALL_USER:
self.actions.append(actions) self.actions.append(actions)
return prediction, ["FAIL"], parsed_responses return prediction, ["FAIL"]
pyautogui_code = parsing_response_to_pyautogui_code( pyautogui_code = parsing_response_to_pyautogui_code(
parsed_response, parsed_response,
@@ -738,7 +741,8 @@ class UITARSAgent:
# Default to FAIL if exceed max steps # Default to FAIL if exceed max steps
actions = ["FAIL"] actions = ["FAIL"]
return prediction, actions, parsed_responses return prediction, actions
@backoff.on_exception( @backoff.on_exception(
backoff.constant, backoff.constant,