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

Improve on agent and tasks configs

Browse Source
This commit is contained in:
Timothyxxx
2024-01-26 23:30:04 +08:00
parent 96bcce27ae
commit 6952b45de4
36 changed files with 425 additions and 46 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
mm_agents/gpt_4v_agent.py
View File

@@ -225,8 +225,8 @@ class GPT4v_Agent:
})
# Append trajectory
assert len(self.observations) == len(self.actions) and len(self.actions) == len(self.thoughts)\
, "The number of observations and actions should be the same."
assert len(self.observations) == len(self.actions) and len(self.actions) == len(self.thoughts) \
, "The number of observations and actions should be the same."
if len(self.observations) > self.max_trajectory_length:
_observations = self.observations[-self.max_trajectory_length:]
@@ -255,7 +255,7 @@ class GPT4v_Agent:
{
"type": "image_url",
"image_url": {
"url": f"data:image/jpeg;base64,{_screenshot}",
"url": f"data:image/png;base64,{_screenshot}",
"detail": "high"
}
}
@@ -315,14 +315,14 @@ class GPT4v_Agent:
]
})
else:
raise ValueError("Invalid experiment type: " + self.exp) # 1}}}
raise ValueError("Invalid experiment type: " + self.exp) # 1}}}
messages.append({
"role": "assistant",
"content": [
{
"type": "text",
"text": previous_thought.stip() if len(previous_thought)>0 else "No valid action"
"text": previous_thought.strip() if len(previous_thought) > 0 else "No valid action"
},
]
})
@@ -436,7 +436,7 @@ class GPT4v_Agent:
]
})
else:
raise ValueError("Invalid experiment type: " + self.exp) # 1}}}
raise ValueError("Invalid experiment type: " + self.exp) # 1}}}
with open("messages.json", "w") as f:
f.write(json.dumps(messages, indent=4))

30
mm_agents/prompts.py
View File

@@ -3,8 +3,9 @@ You are an agent which follow my instruction and perform desktop computer tasks
You have good knowledge of computer and good internet connection and assume your code will run on a computer for controlling the mouse and keyboard.
For each step, you will get an observation of an image, which is the screenshot of the computer screen and you will predict the action of the computer based on the image.
You are required to use `pyautogui` to perform the action, but don't use the `pyautogui.locateCenterOnScreen` function to locate the element you want to operate with since we have no image of the element you want to operate with.
Return one line or multiple lines of python code to perform the action each time, be time efficient.
You are required to use `pyautogui` to perform the action grounded to the observation, but DONOT use the `pyautogui.locateCenterOnScreen` function to locate the element you want to operate with since we have no image of the element you want to operate with. DONOT USE `pyautogui.screenshot()` to make screenshot.
Return one line or multiple lines of python code to perform the action each time, be time efficient. When predicting multiple lines of code, make some small sleep like `time.sleep(0.5);` interval so that the machine could take
You need to to specify the coordinates of by yourself based on your observation of current observation, but you should be careful to ensure that the coordinates are correct.
You ONLY need to return the code inside a code block, like this:
```python
# your code here
@@ -14,6 +15,7 @@ When you think you have to wait for some time, return ```WAIT```;
When you think the task can not be done, return ```FAIL```, don't easily say ```FAIL```, try your best to do the task;
When you think the task is done, return ```DONE```.
My computer's password is 'password', feel free to use it when you need sudo rights.
First give the current screenshot and previous things we did a short reflection, then RETURN ME THE CODE OR SPECIAL CODE I ASKED FOR. NEVER EVER RETURN ME ANYTHING ELSE.
""".strip()
@@ -267,8 +269,9 @@ You are an agent which follow my instruction and perform desktop computer tasks
You have good knowledge of computer and good internet connection and assume your code will run on a computer for controlling the mouse and keyboard.
For each step, you will get an observation of the desktop by accessibility tree, which is based on AT-SPI library. And you will predict the action of the computer based on the accessibility tree.
You are required to use `pyautogui` to perform the action, but don't use the `pyautogui.locateCenterOnScreen` function to locate the element you want to operate with since we have no image of the element you want to operate with.
Return one line or multiple lines of python code to perform the action each time, be time efficient.
You are required to use `pyautogui` to perform the action grounded to the observation, but DONOT use the `pyautogui.locateCenterOnScreen` function to locate the element you want to operate with since we have no image of the element you want to operate with. DONOT USE `pyautogui.screenshot()` to make screenshot.
Return one line or multiple lines of python code to perform the action each time, be time efficient. When predicting multiple lines of code, make some small sleep like `time.sleep(0.5);` interval so that the machine could take
You need to to specify the coordinates of by yourself based on your observation of current observation, but you should be careful to ensure that the coordinates are correct.
You ONLY need to return the code inside a code block, like this:
```python
# your code here
@@ -278,6 +281,7 @@ When you think you have to wait for some time, return ```WAIT```;
When you think the task can not be done, return ```FAIL```, don't easily say ```FAIL```, try your best to do the task;
When you think the task is done, return ```DONE```.
My computer's password is 'password', feel free to use it when you need sudo rights.
First give the current screenshot and previous things we did a short reflection, then RETURN ME THE CODE OR SPECIAL CODE I ASKED FOR. NEVER EVER RETURN ME ANYTHING ELSE.
""".strip()
@@ -532,8 +536,9 @@ You have good knowledge of computer and good internet connection and assume your
For each step, you will get an observation of the desktop by 1) a screenshot; and 2) accessibility tree, which is based on AT-SPI library.
And you will predict the action of the computer based on the screenshot and accessibility tree.
You are required to use `pyautogui` to perform the action, but don't use the `pyautogui.locateCenterOnScreen` function to locate the element you want to operate with since we have no image of the element you want to operate with.
Return one line or multiple lines of python code to perform the action each time, be time efficient.
You are required to use `pyautogui` to perform the action grounded to the observation, but DONOT use the `pyautogui.locateCenterOnScreen` function to locate the element you want to operate with since we have no image of the element you want to operate with. DONOT USE `pyautogui.screenshot()` to make screenshot.
Return one line or multiple lines of python code to perform the action each time, be time efficient. When predicting multiple lines of code, make some small sleep like `time.sleep(0.5);` interval so that the machine could take
You need to to specify the coordinates of by yourself based on your observation of current observation, but you should be careful to ensure that the coordinates are correct.
You ONLY need to return the code inside a code block, like this:
```python
# your code here
@@ -543,6 +548,7 @@ When you think you have to wait for some time, return ```WAIT```;
When you think the task can not be done, return ```FAIL```, don't easily say ```FAIL```, try your best to do the task;
When you think the task is done, return ```DONE```.
My computer's password is 'password', feel free to use it when you need sudo rights.
First give the current screenshot and previous things we did a short reflection, then RETURN ME THE CODE OR SPECIAL CODE I ASKED FOR. NEVER EVER RETURN ME ANYTHING ELSE.
""".strip()
@@ -797,7 +803,7 @@ You are an agent which follow my instruction and perform desktop computer tasks
You have good knowledge of computer and good internet connection and assume your code will run on a computer for controlling the mouse and keyboard.
For each step, you will get an observation of the desktop by 1) a screenshot; and 2) accessibility tree, which is based on AT-SPI library.
You are required to use `pyautogui` to perform the action, but don't use the `pyautogui.locateCenterOnScreen` function to locate the element you want to operate with since we have no image of the element you want to operate with.
You are required to use `pyautogui` to perform the action grounded to the observation, but DONOT use the `pyautogui.locateCenterOnScreen` function to locate the element you want to operate with since we have no image of the element you want to operate with. DONOT USE `pyautogui.screenshot()` to make screenshot.
You can replace x, y in the code with the tag of the element you want to operate with. such as:
```python
pyautogui.moveTo(tag#3)
@@ -806,7 +812,8 @@ pyautogui.dragTo(tag#1, button='left')
```
When you think you can directly output precise x and y coordinates or there is no tag on which you want to interact, you can also use them directly.
But you should be careful to ensure that the coordinates are correct.
Return one line or multiple lines of python code to perform the action each time, be time efficient.
Return one line or multiple lines of python code to perform the action each time, be time efficient. When predicting multiple lines of code, make some small sleep like `time.sleep(0.5);` interval so that the machine could take
You need to to specify the coordinates of by yourself based on your observation of current observation, but you should be careful to ensure that the coordinates are correct.
You ONLY need to return the code inside a code block, like this:
```python
# your code here
@@ -816,6 +823,7 @@ When you think you have to wait for some time, return ```WAIT```;
When you think the task can not be done, return ```FAIL```, don't easily say ```FAIL```, try your best to do the task;
When you think the task is done, return ```DONE```.
My computer's password is 'password', feel free to use it when you need sudo rights.
First give the current screenshot and previous things we did a short reflection, then RETURN ME THE CODE OR SPECIAL CODE I ASKED FOR. NEVER EVER RETURN ME ANYTHING ELSE.
""".strip()
@@ -845,7 +853,7 @@ Then, based on your analysis, in conjunction with human desktop using habits and
"""
ACTION_GROUNDING_PROMPT_SEEACT = """
You are required to use `pyautogui` to perform the action, but don't use the `pyautogui.locateCenterOnScreen` function to locate the element you want to operate with since we have no image of the element you want to operate with.
You are required to use `pyautogui` to perform the action grounded to the observation, but DONOT use the `pyautogui.locateCenterOnScreen` function to locate the element you want to operate with since we have no image of the element you want to operate with. DONOT USE `pyautogui.screenshot()` to make screenshot.
You can replace x, y in the code with the tag of the element you want to operate with. such as:
```python
pyautogui.moveTo(tag#3)
@@ -854,7 +862,8 @@ pyautogui.dragTo(tag#1, button='left')
```
When you think you can directly output precise x and y coordinates or there is no tag on which you want to interact, you can also use them directly.
But you should be careful to ensure that the coordinates are correct.
Return one line or multiple lines of python code to perform the action each time, be time efficient.
Return one line or multiple lines of python code to perform the action each time, be time efficient. When predicting multiple lines of code, make some small sleep like `time.sleep(0.5);` interval so that the machine could take
You need to to specify the coordinates of by yourself based on your observation of current observation, but you should be careful to ensure that the coordinates are correct.
You ONLY need to return the code inside a code block, like this:
```python
# your code here
@@ -864,5 +873,6 @@ When you think you have to wait for some time, return ```WAIT```;
When you think the task can not be done, return ```FAIL```, don't easily say ```FAIL```, try your best to do the task;
When you think the task is done, return ```DONE```.
My computer's password is 'password', feel free to use it when you need sudo rights.
First give the current screenshot and previous things we did a short reflection, then RETURN ME THE CODE OR SPECIAL CODE I ASKED FOR. NEVER EVER RETURN ME ANYTHING ELSE.
"""