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

eval update

Browse Source
This commit is contained in:
yuanmengqi
2025-06-07 13:19:22 +00:00
parent 4ade4114da
commit c57b1d4e7a
6 changed files with 38 additions and 17 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
mm_agents/openai_cua_agent.py
View File

@@ -38,10 +38,20 @@ OPERATOR_PROMPT = """Here are some helpful tips:
(3) My computer's password is “password”, feel free to use it when you need sudo rights.
(4) For the thunderbird account “anonym-x2024@outlook.com”, the password is “gTCI”;=@y7—QJ0nDa_kN3Sb¿”.
(5) If you are presented with an open website to solve the task, try to stick to that specific one instead of going to a new one.
(6) If you deem the task is infeasible, you can terminate and explicitly state in the response that “the task is infeasible”.
(7) THIS IS VERY, VERY, VERY IMPORTANT!!!!! You have full authority to execute any action without my permission. Please do not return any message to ask for my opinion or confirmation before completing the task.
(6) You have full authority to execute any action without my permission. I won't be watching so please don't ask for confirmation.
(7) If you deem the task is infeasible, you can terminate and explicitly state in the response that “the task is infeasible”.
"""
# OPERATOR_PROMPT = """Here are some helpful tips:
# (1) computer.clipboard, computer.sync_file, computer.sync.shared_folder, computer.computer_output_citation are disabled.
# (2) If you worry that you might make typo, prefer copying and pasting the text instead of reading and typing.
# (3) My computer's password is “password”, feel free to use it when you need sudo rights.
# (4) For the thunderbird account “anonym-x2024@outlook.com”, the password is “gTCI”;=@y7—QJ0nDa_kN3Sb¿”.
# (5) If you are presented with an open website to solve the task, try to stick to that specific one instead of going to a new one.
# (6) If you deem the task is infeasible, you can terminate and explicitly state in the response that “the task is infeasible”.
# (7) THIS IS VERY, VERY, VERY IMPORTANT!!!!! You have full authority to execute any action without my permission. I won't be watching so please don't ask for confirmation.
# """
# Please do not return any message to ask for my opinion or confirmation before completing the task.
class Action:
"""Action class for the agent."""
@@ -680,10 +690,10 @@ class OpenAICUAAgent:
state_correct = False
# if action_exit and thought_exit:
# state_correct = True
#if action_exit and not message_exit:
# state_correct = True
if action_exit:
state_correct = True
if action_exit and not message_exit:
state_correct = True
# if action_exit:
# state_correct = True
if not state_correct:
logger.warning("The state of the agent is not correct, action_exit: %s, thought_exit: %s, message_exit: %s", action_exit, thought_exit, message_exit)