messages
listlengths 2
200
| images
listlengths 1
3
| custom_id
int64 0
66.2k
⌀ |
|---|---|---|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Search for hotels in Miami<image>Thought: Scroll down to view more hotel options listed for Miami.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>Thought: Scroll down to view more hotel options and pricing in Miami.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Click on the TripAdvisor link to explore hotel options for Miami in 2022.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [284, 391]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/8723752525844759828_9.png",
"training_data/AITW/aitw_images/general/8723752525844759828_10.png",
"training_data/AITW/aitw_images/general/8723752525844759828_11.png"
] | 2,000
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Search for hotels in Miami<image>Thought: Scroll down to view more hotel options and pricing in Miami.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>Thought: Click on the TripAdvisor link to explore hotel options for Miami in 2022.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [284, 391]}<image>",
"role": "user"
},
{
"content": "Thought: The task is marked as complete because the user successfully searched for hotels in Miami using the provided interface.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/8723752525844759828_10.png",
"training_data/AITW/aitw_images/general/8723752525844759828_11.png",
"training_data/AITW/aitw_images/general/8723752525844759828_12.png"
] | 2,001
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Check the settings for the Netflix app<image>",
"role": "user"
},
{
"content": "Thought: You should have clicked the Netflix app icon instead of PLAY to access the settings.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [225, 467]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/15076535410203209250_0.png"
] | 2,002
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Check the settings for the Netflix app<image>Thought: You should have clicked the Netflix app icon instead of PLAY to access the settings.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [225, 467]}<image>",
"role": "user"
},
{
"content": "Thought: Click the MAGNIFYING GLASS to access search functionality for finding the Netflix app settings.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [210, 35]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/15076535410203209250_0.png",
"training_data/AITW/aitw_images/general/15076535410203209250_1.png"
] | 2,003
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Check the settings for the Netflix app<image>Thought: You should have clicked the Netflix app icon instead of PLAY to access the settings.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [225, 467]}<image>Thought: Click the MAGNIFYING GLASS to access search functionality for finding the Netflix app settings.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [210, 35]}<image>",
"role": "user"
},
{
"content": "Thought: Click the X to close the current search for apps and return to the previous screen.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [254, 40]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/15076535410203209250_0.png",
"training_data/AITW/aitw_images/general/15076535410203209250_1.png",
"training_data/AITW/aitw_images/general/15076535410203209250_2.png"
] | 2,004
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Check the settings for the Netflix app<image>Thought: Click the MAGNIFYING GLASS to access search functionality for finding the Netflix app settings.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [210, 35]}<image>Thought: Click the X to close the current search for apps and return to the previous screen.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [254, 40]}<image>",
"role": "user"
},
{
"content": "Thought: Type \"netflix\" in the search bar to find the app settings.\nAction: {\"action\": \"TYPE\", \"value\": \"netflix\", \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/15076535410203209250_1.png",
"training_data/AITW/aitw_images/general/15076535410203209250_2.png",
"training_data/AITW/aitw_images/general/15076535410203209250_3.png"
] | 2,005
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Check the settings for the Netflix app<image>Thought: Click the X to close the current search for apps and return to the previous screen.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [254, 40]}<image>Thought: Type \"netflix\" in the search bar to find the app settings.\nAction: {\"action\": \"TYPE\", \"value\": \"netflix\", \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Click X to exit the search for Netflix and return to the main screen.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [190, 105]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/15076535410203209250_2.png",
"training_data/AITW/aitw_images/general/15076535410203209250_3.png",
"training_data/AITW/aitw_images/general/15076535410203209250_4.png"
] | 2,006
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Check the settings for the Netflix app<image>Thought: Type \"netflix\" in the search bar to find the app settings.\nAction: {\"action\": \"TYPE\", \"value\": \"netflix\", \"position\": null}<image>Thought: Click X to exit the search for Netflix and return to the main screen.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [190, 105]}<image>",
"role": "user"
},
{
"content": "Thought: You checked the settings for the Netflix app after searching for it. The search results confirmed that you found the correct app to access its settings.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/15076535410203209250_3.png",
"training_data/AITW/aitw_images/general/15076535410203209250_4.png",
"training_data/AITW/aitw_images/general/15076535410203209250_5.png"
] | 2,007
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Check the settings for the Google Play Books app<image>",
"role": "user"
},
{
"content": "Thought: Press home to return to the main screen and access apps for checking Google Play Books settings.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/17194776140386962102_0.png"
] | 2,008
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Check the settings for the Google Play Books app<image>Thought: Press home to return to the main screen and access apps for checking Google Play Books settings.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Your task is to confirm you accessed the settings for Google Play Books successfully. The complete status indicates that you have verified the settings.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/17194776140386962102_0.png",
"training_data/AITW/aitw_images/general/17194776140386962102_1.png"
] | 2,009
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What's the latest video from GameSpot Trailers?<image>",
"role": "user"
},
{
"content": "Thought: Click PLAY to start watching the latest video from GameSpot Trailers on YouTube.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [607, 933]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4795384570017677048_0.png"
] | 2,010
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What's the latest video from GameSpot Trailers?<image>Thought: Click PLAY to start watching the latest video from GameSpot Trailers on YouTube.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [607, 933]}<image>",
"role": "user"
},
{
"content": "Thought: Click on the downward arrow to access more video options or the latest content.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [53, 91]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4795384570017677048_0.png",
"training_data/AITW/aitw_images/general/4795384570017677048_1.png"
] | 2,011
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What's the latest video from GameSpot Trailers?<image>Thought: Click PLAY to start watching the latest video from GameSpot Trailers on YouTube.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [607, 933]}<image>Thought: Click on the downward arrow to access more video options or the latest content.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [53, 91]}<image>",
"role": "user"
},
{
"content": "Thought: Click the magnifying glass to access the search feature for finding GameSpot Trailers' latest video.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [599, 81]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4795384570017677048_0.png",
"training_data/AITW/aitw_images/general/4795384570017677048_1.png",
"training_data/AITW/aitw_images/general/4795384570017677048_2.png"
] | 2,012
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What's the latest video from GameSpot Trailers?<image>Thought: Click on the downward arrow to access more video options or the latest content.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [53, 91]}<image>Thought: Click the magnifying glass to access the search feature for finding GameSpot Trailers' latest video.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [599, 81]}<image>",
"role": "user"
},
{
"content": "Thought: Click on GameSpot Trailers to access their latest video content.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [215, 421]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4795384570017677048_1.png",
"training_data/AITW/aitw_images/general/4795384570017677048_2.png",
"training_data/AITW/aitw_images/general/4795384570017677048_3.png"
] | 2,013
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What's the latest video from GameSpot Trailers?<image>Thought: Click the magnifying glass to access the search feature for finding GameSpot Trailers' latest video.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [599, 81]}<image>Thought: Click on GameSpot Trailers to access their latest video content.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [215, 421]}<image>",
"role": "user"
},
{
"content": "Thought: Scroll down to view more content and find the latest video from GameSpot Trailers.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4795384570017677048_2.png",
"training_data/AITW/aitw_images/general/4795384570017677048_3.png",
"training_data/AITW/aitw_images/general/4795384570017677048_4.png"
] | 2,014
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What's the latest video from GameSpot Trailers?<image>Thought: Click on GameSpot Trailers to access their latest video content.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [215, 421]}<image>Thought: Scroll down to view more content and find the latest video from GameSpot Trailers.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Click on the \"CHARLES\" thumbnail to access the latest video from GameSpot Trailers.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [352, 1008]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4795384570017677048_3.png",
"training_data/AITW/aitw_images/general/4795384570017677048_4.png",
"training_data/AITW/aitw_images/general/4795384570017677048_5.png"
] | 2,015
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What's the latest video from GameSpot Trailers?<image>Thought: Scroll down to view more content and find the latest video from GameSpot Trailers.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>Thought: Click on the \"CHARLES\" thumbnail to access the latest video from GameSpot Trailers.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [352, 1008]}<image>",
"role": "user"
},
{
"content": "Thought: Click \"la\" to complete your search for GameSpot Trailers content quickly and efficiently.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [351, 248]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4795384570017677048_4.png",
"training_data/AITW/aitw_images/general/4795384570017677048_5.png",
"training_data/AITW/aitw_images/general/4795384570017677048_6.png"
] | 2,016
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What's the latest video from GameSpot Trailers?<image>Thought: Click on the \"CHARLES\" thumbnail to access the latest video from GameSpot Trailers.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [352, 1008]}<image>Thought: Click \"la\" to complete your search for GameSpot Trailers content quickly and efficiently.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [351, 248]}<image>",
"role": "user"
},
{
"content": "Thought: Pause the ad to return to watching the latest video from GameSpot Trailers without interruptions.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [368, 245]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4795384570017677048_5.png",
"training_data/AITW/aitw_images/general/4795384570017677048_6.png",
"training_data/AITW/aitw_images/general/4795384570017677048_7.png"
] | 2,017
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What's the latest video from GameSpot Trailers?<image>Thought: Click \"la\" to complete your search for GameSpot Trailers content quickly and efficiently.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [351, 248]}<image>Thought: Pause the ad to return to watching the latest video from GameSpot Trailers without interruptions.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [368, 245]}<image>",
"role": "user"
},
{
"content": "Thought: The task is complete because the latest video from GameSpot Trailers has been accessed and viewed.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4795384570017677048_6.png",
"training_data/AITW/aitw_images/general/4795384570017677048_7.png",
"training_data/AITW/aitw_images/general/4795384570017677048_8.png"
] | 2,018
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Search for the best burger restaurants on Maps<image>",
"role": "user"
},
{
"content": "Thought: Press home to exit the current app and return to the main screen for easier navigation.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/3747125767618087904_0.png"
] | 2,019
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Search for the best burger restaurants on Maps<image>Thought: Press home to exit the current app and return to the main screen for easier navigation.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Click the NAV BAR CIRCLE to open the main menu and access apps for searching burger restaurants.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [246, 978]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/3747125767618087904_0.png",
"training_data/AITW/aitw_images/general/3747125767618087904_1.png"
] | 2,020
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Search for the best burger restaurants on Maps<image>Thought: Press home to exit the current app and return to the main screen for easier navigation.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>Thought: Click the NAV BAR CIRCLE to open the main menu and access apps for searching burger restaurants.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [246, 978]}<image>",
"role": "user"
},
{
"content": "Thought: Type \"best burger restaurants\" in the search bar to find relevant results on Maps.\nAction: {\"action\": \"TYPE\", \"value\": \"best burger restaurants\", \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/3747125767618087904_0.png",
"training_data/AITW/aitw_images/general/3747125767618087904_1.png",
"training_data/AITW/aitw_images/general/3747125767618087904_2.png"
] | 2,021
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Search for the best burger restaurants on Maps<image>Thought: Click the NAV BAR CIRCLE to open the main menu and access apps for searching burger restaurants.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [246, 978]}<image>Thought: Type \"best burger restaurants\" in the search bar to find relevant results on Maps.\nAction: {\"action\": \"TYPE\", \"value\": \"best burger restaurants\", \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Type the search query \"best burger restaurant\" to find relevant results on Google.\nAction: {\"action\": \"TYPE\", \"value\": \"\", \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/3747125767618087904_1.png",
"training_data/AITW/aitw_images/general/3747125767618087904_2.png",
"training_data/AITW/aitw_images/general/3747125767618087904_3.png"
] | 2,022
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Search for the best burger restaurants on Maps<image>Thought: Type \"best burger restaurants\" in the search bar to find relevant results on Maps.\nAction: {\"action\": \"TYPE\", \"value\": \"best burger restaurants\", \"position\": null}<image>Thought: Type the search query \"best burger restaurant\" to find relevant results on Google.\nAction: {\"action\": \"TYPE\", \"value\": \"\", \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Click on \"best burger restaurants near me\" to find local options quickly.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [322, 143]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/3747125767618087904_2.png",
"training_data/AITW/aitw_images/general/3747125767618087904_3.png",
"training_data/AITW/aitw_images/general/3747125767618087904_4.png"
] | 2,023
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Search for the best burger restaurants on Maps<image>Thought: Type the search query \"best burger restaurant\" to find relevant results on Google.\nAction: {\"action\": \"TYPE\", \"value\": \"\", \"position\": null}<image>Thought: Click on \"best burger restaurants near me\" to find local options quickly.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [322, 143]}<image>",
"role": "user"
},
{
"content": "Thought: Click on LOCATION to see nearby burger restaurants that match my search criteria.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [289, 566]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/3747125767618087904_3.png",
"training_data/AITW/aitw_images/general/3747125767618087904_4.png",
"training_data/AITW/aitw_images/general/3747125767618087904_5.png"
] | 2,024
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Search for the best burger restaurants on Maps<image>Thought: Click on \"best burger restaurants near me\" to find local options quickly.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [322, 143]}<image>Thought: Click on LOCATION to see nearby burger restaurants that match my search criteria.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [289, 566]}<image>",
"role": "user"
},
{
"content": "Thought: You have successfully searched for \"best burger restaurants near me\" on Maps. The map now shows various burger restaurants in the desired area.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/3747125767618087904_4.png",
"training_data/AITW/aitw_images/general/3747125767618087904_5.png",
"training_data/AITW/aitw_images/general/3747125767618087904_6.png"
] | 2,025
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: How much does the HisenseTV cost?<image>",
"role": "user"
},
{
"content": "Thought: Open the Google app to search for Hisense TV prices and mark the task as complete.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/7993026574296149770_0.png"
] | 2,026
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What is the speed of light?<image>",
"role": "user"
},
{
"content": "Thought: Press home because the Google Photos app is prompting for an update, preventing further use of the app.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/6816641115564678333_0.png"
] | 2,027
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What is the speed of light?<image>Thought: Press home because the Google Photos app is prompting for an update, preventing further use of the app.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Click the Google Chrome icon to open a browser for searching.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [387, 920]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/6816641115564678333_0.png",
"training_data/AITW/aitw_images/general/6816641115564678333_1.png"
] | 2,028
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What is the speed of light?<image>Thought: Press home because the Google Photos app is prompting for an update, preventing further use of the app.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>Thought: Click the Google Chrome icon to open a browser for searching.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [387, 920]}<image>",
"role": "user"
},
{
"content": "Thought: Status task impossible: The user is searching for the speed of light but is on a restaurant search page.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [240, 65]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/6816641115564678333_0.png",
"training_data/AITW/aitw_images/general/6816641115564678333_1.png",
"training_data/AITW/aitw_images/general/6816641115564678333_2.png"
] | 2,029
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What is the speed of light?<image>Thought: Click the Google Chrome icon to open a browser for searching.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [387, 920]}<image>Thought: Status task impossible: The user is searching for the speed of light but is on a restaurant search page.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [240, 65]}<image>",
"role": "user"
},
{
"content": "Thought: Click the URL to alter the search and potentially find relevant information on light speed.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [240, 65]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/6816641115564678333_1.png",
"training_data/AITW/aitw_images/general/6816641115564678333_2.png",
"training_data/AITW/aitw_images/general/6816641115564678333_3.png"
] | 2,030
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What is the speed of light?<image>Thought: Status task impossible: The user is searching for the speed of light but is on a restaurant search page.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [240, 65]}<image>Thought: Click the URL to alter the search and potentially find relevant information on light speed.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [240, 65]}<image>",
"role": "user"
},
{
"content": "Thought: You likely mistyped your query since you're searching for information about the speed of light, not restaurants.\nAction: {\"action\": \"TYPE\", \"value\": \"speed of light\", \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/6816641115564678333_2.png",
"training_data/AITW/aitw_images/general/6816641115564678333_3.png",
"training_data/AITW/aitw_images/general/6816641115564678333_4.png"
] | 2,031
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What is the speed of light?<image>Thought: Click the URL to alter the search and potentially find relevant information on light speed.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [240, 65]}<image>Thought: You likely mistyped your query since you're searching for information about the speed of light, not restaurants.\nAction: {\"action\": \"TYPE\", \"value\": \"speed of light\", \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Click the 'X' to clear the current search query, so I can start a new search for \"speed of light.\"\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [506, 66]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/6816641115564678333_3.png",
"training_data/AITW/aitw_images/general/6816641115564678333_4.png",
"training_data/AITW/aitw_images/general/6816641115564678333_5.png"
] | 2,032
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What is the speed of light?<image>Thought: You likely mistyped your query since you're searching for information about the speed of light, not restaurants.\nAction: {\"action\": \"TYPE\", \"value\": \"speed of light\", \"position\": null}<image>Thought: Click the 'X' to clear the current search query, so I can start a new search for \"speed of light.\"\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [506, 66]}<image>",
"role": "user"
},
{
"content": "Thought: You type \"speed of light\" in the search box to find relevant information about it.\nAction: {\"action\": \"TYPE\", \"value\": \"speed of light\", \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/6816641115564678333_4.png",
"training_data/AITW/aitw_images/general/6816641115564678333_5.png",
"training_data/AITW/aitw_images/general/6816641115564678333_6.png"
] | 2,033
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What is the speed of light?<image>Thought: Click the 'X' to clear the current search query, so I can start a new search for \"speed of light.\"\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [506, 66]}<image>Thought: You type \"speed of light\" in the search box to find relevant information about it.\nAction: {\"action\": \"TYPE\", \"value\": \"speed of light\", \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Click on the result showing 'speed of light = 299792458 m / s' to find the answer.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [262, 194]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/6816641115564678333_5.png",
"training_data/AITW/aitw_images/general/6816641115564678333_6.png",
"training_data/AITW/aitw_images/general/6816641115564678333_7.png"
] | 2,034
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What is the speed of light?<image>Thought: You type \"speed of light\" in the search box to find relevant information about it.\nAction: {\"action\": \"TYPE\", \"value\": \"speed of light\", \"position\": null}<image>Thought: Click on the result showing 'speed of light = 299792458 m / s' to find the answer.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [262, 194]}<image>",
"role": "user"
},
{
"content": "Thought: You see the search results regarding the speed of light. You confirm that you found the information you needed.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/6816641115564678333_6.png",
"training_data/AITW/aitw_images/general/6816641115564678333_7.png",
"training_data/AITW/aitw_images/general/6816641115564678333_8.png"
] | 2,035
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: How much does the Vizio TV cost?<image>",
"role": "user"
},
{
"content": "Thought: Press home to exit the current screen and access the main menu for further navigation.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/1104668595957467721_0.png"
] | 2,036
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: How much does the Vizio TV cost?<image>Thought: Press home to exit the current screen and access the main menu for further navigation.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Click PLAY to open the app for shopping or streaming that may provide Vizio TV prices.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [260, 538]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/1104668595957467721_0.png",
"training_data/AITW/aitw_images/general/1104668595957467721_1.png"
] | 2,037
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: How much does the Vizio TV cost?<image>Thought: Press home to exit the current screen and access the main menu for further navigation.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>Thought: Click PLAY to open the app for shopping or streaming that may provide Vizio TV prices.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [260, 538]}<image>",
"role": "user"
},
{
"content": "Thought: You clicked on g.trycaviar.com to access restaurant information, possibly looking for Vizio TV pricing details.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [171, 75]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/1104668595957467721_0.png",
"training_data/AITW/aitw_images/general/1104668595957467721_1.png",
"training_data/AITW/aitw_images/general/1104668595957467721_2.png"
] | 2,038
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: How much does the Vizio TV cost?<image>Thought: Click PLAY to open the app for shopping or streaming that may provide Vizio TV prices.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [260, 538]}<image>Thought: You clicked on g.trycaviar.com to access restaurant information, possibly looking for Vizio TV pricing details.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [171, 75]}<image>",
"role": "user"
},
{
"content": "Thought: Type the query in the search bar to find pricing information for the Vizio TV.\nAction: {\"action\": \"TYPE\", \"value\": \"How much does the Vizio TV cost?\", \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/1104668595957467721_1.png",
"training_data/AITW/aitw_images/general/1104668595957467721_2.png",
"training_data/AITW/aitw_images/general/1104668595957467721_3.png"
] | 2,039
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: How much does the Vizio TV cost?<image>Thought: You clicked on g.trycaviar.com to access restaurant information, possibly looking for Vizio TV pricing details.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [171, 75]}<image>Thought: Type the query in the search bar to find pricing information for the Vizio TV.\nAction: {\"action\": \"TYPE\", \"value\": \"How much does the Vizio TV cost?\", \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Click the suggestion \"How much does the Vizio TV ...\" to get specific pricing information.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [193, 146]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/1104668595957467721_2.png",
"training_data/AITW/aitw_images/general/1104668595957467721_3.png",
"training_data/AITW/aitw_images/general/1104668595957467721_4.png"
] | 2,040
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: How much does the Vizio TV cost?<image>Thought: Type the query in the search bar to find pricing information for the Vizio TV.\nAction: {\"action\": \"TYPE\", \"value\": \"How much does the Vizio TV cost?\", \"position\": null}<image>Thought: Click the suggestion \"How much does the Vizio TV ...\" to get specific pricing information.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [193, 146]}<image>",
"role": "user"
},
{
"content": "Thought: Recognize that the search query displays TV prices, including options for Vizio models. Confirm that I have found the desired information after assessing results.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/1104668595957467721_3.png",
"training_data/AITW/aitw_images/general/1104668595957467721_4.png",
"training_data/AITW/aitw_images/general/1104668595957467721_5.png"
] | 2,041
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What's the news about the US president?<image>",
"role": "user"
},
{
"content": "Thought: Press home to exit the Settings menu and return to the main screen for news access.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/8518150215196527742_0.png"
] | 2,042
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What's the news about the US president?<image>Thought: Press home to exit the Settings menu and return to the main screen for news access.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Click on Photos to view images related to current events, possibly including news about the US president.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [168, 467]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/8518150215196527742_0.png",
"training_data/AITW/aitw_images/general/8518150215196527742_1.png"
] | 2,043
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What's the news about the US president?<image>Thought: Press home to exit the Settings menu and return to the main screen for news access.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>Thought: Click on Photos to view images related to current events, possibly including news about the US president.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [168, 467]}<image>",
"role": "user"
},
{
"content": "Thought: The user likely clicked on the Google search bar to enter their query about the US president.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [146, 26]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/8518150215196527742_0.png",
"training_data/AITW/aitw_images/general/8518150215196527742_1.png",
"training_data/AITW/aitw_images/general/8518150215196527742_2.png"
] | 2,044
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What's the news about the US president?<image>Thought: Click on Photos to view images related to current events, possibly including news about the US president.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [168, 467]}<image>Thought: The user likely clicked on the Google search bar to enter their query about the US president.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [146, 26]}<image>",
"role": "user"
},
{
"content": "Thought: Type in the search bar to find articles related to the US president.\nAction: {\"action\": \"TYPE\", \"value\": \"What's the news about the US president?\", \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/8518150215196527742_1.png",
"training_data/AITW/aitw_images/general/8518150215196527742_2.png",
"training_data/AITW/aitw_images/general/8518150215196527742_3.png"
] | 2,045
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What's the news about the US president?<image>Thought: The user likely clicked on the Google search bar to enter their query about the US president.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [146, 26]}<image>Thought: Type in the search bar to find articles related to the US president.\nAction: {\"action\": \"TYPE\", \"value\": \"What's the news about the US president?\", \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Click the search bar to initiate a query for news about the US president.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [222, 71]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/8518150215196527742_2.png",
"training_data/AITW/aitw_images/general/8518150215196527742_3.png",
"training_data/AITW/aitw_images/general/8518150215196527742_4.png"
] | 2,046
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What's the news about the US president?<image>Thought: Type in the search bar to find articles related to the US president.\nAction: {\"action\": \"TYPE\", \"value\": \"What's the news about the US president?\", \"position\": null}<image>Thought: Click the search bar to initiate a query for news about the US president.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [222, 71]}<image>",
"role": "user"
},
{
"content": "Thought: Click on the \"Joe Biden news\" link to access specific updates about the US president.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [58, 184]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/8518150215196527742_3.png",
"training_data/AITW/aitw_images/general/8518150215196527742_4.png",
"training_data/AITW/aitw_images/general/8518150215196527742_5.png"
] | 2,047
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What's the news about the US president?<image>Thought: Click the search bar to initiate a query for news about the US president.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [222, 71]}<image>Thought: Click on the \"Joe Biden news\" link to access specific updates about the US president.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [58, 184]}<image>",
"role": "user"
},
{
"content": "Thought: You likely opened the CNN politics page to find the latest news about the US president, Joe Biden. After scrolling through the articles and gathering information, you marked your task as complete since you found relevant news updates.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/8518150215196527742_4.png",
"training_data/AITW/aitw_images/general/8518150215196527742_5.png",
"training_data/AITW/aitw_images/general/8518150215196527742_6.png"
] | 2,048
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Set an alarm for 12pm<image>",
"role": "user"
},
{
"content": "Thought: Scroll down to find the clock app icon on the home screen.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/6271175289399993355_0.png"
] | 2,049
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Set an alarm for 12pm<image>Thought: Scroll down to find the clock app icon on the home screen.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Click GOOGLE to open the app that allows me to set an alarm.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [32, 341]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/6271175289399993355_0.png",
"training_data/AITW/aitw_images/general/6271175289399993355_1.png"
] | 2,050
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Set an alarm for 12pm<image>Thought: Scroll down to find the clock app icon on the home screen.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>Thought: Click GOOGLE to open the app that allows me to set an alarm.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [32, 341]}<image>",
"role": "user"
},
{
"content": "Thought: Click the search bar to enter an alarm app.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [184, 70]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/6271175289399993355_0.png",
"training_data/AITW/aitw_images/general/6271175289399993355_1.png",
"training_data/AITW/aitw_images/general/6271175289399993355_2.png"
] | 2,051
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Set an alarm for 12pm<image>Thought: Click GOOGLE to open the app that allows me to set an alarm.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [32, 341]}<image>Thought: Click the search bar to enter an alarm app.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [184, 70]}<image>",
"role": "user"
},
{
"content": "Thought: You likely clicked X to close the search results for flights, as it's not related to setting an alarm.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [251, 28]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/6271175289399993355_1.png",
"training_data/AITW/aitw_images/general/6271175289399993355_2.png",
"training_data/AITW/aitw_images/general/6271175289399993355_3.png"
] | 2,052
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Set an alarm for 12pm<image>Thought: Click the search bar to enter an alarm app.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [184, 70]}<image>Thought: You likely clicked X to close the search results for flights, as it's not related to setting an alarm.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [251, 28]}<image>",
"role": "user"
},
{
"content": "Thought: Press home to exit the current screen and access the main menu for setting an alarm.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/6271175289399993355_2.png",
"training_data/AITW/aitw_images/general/6271175289399993355_3.png",
"training_data/AITW/aitw_images/general/6271175289399993355_4.png"
] | 2,053
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Set an alarm for 12pm<image>Thought: You likely clicked X to close the search results for flights, as it's not related to setting an alarm.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [251, 28]}<image>Thought: Press home to exit the current screen and access the main menu for setting an alarm.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Scroll down to find the alarm app or widget on the home screen.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/6271175289399993355_3.png",
"training_data/AITW/aitw_images/general/6271175289399993355_4.png",
"training_data/AITW/aitw_images/general/6271175289399993355_5.png"
] | 2,054
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Set an alarm for 12pm<image>Thought: Press home to exit the current screen and access the main menu for setting an alarm.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>Thought: Scroll down to find the alarm app or widget on the home screen.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Click on the Clock app icon to access the alarm settings.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [216, 177]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/6271175289399993355_4.png",
"training_data/AITW/aitw_images/general/6271175289399993355_5.png",
"training_data/AITW/aitw_images/general/6271175289399993355_6.png"
] | 2,055
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Set an alarm for 12pm<image>Thought: Scroll down to find the alarm app or widget on the home screen.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>Thought: Click on the Clock app icon to access the alarm settings.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [216, 177]}<image>",
"role": "user"
},
{
"content": "Thought: Click PLUS to add a new alarm.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [141, 491]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/6271175289399993355_5.png",
"training_data/AITW/aitw_images/general/6271175289399993355_6.png",
"training_data/AITW/aitw_images/general/6271175289399993355_7.png"
] | 2,056
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Set an alarm for 12pm<image>Thought: Click on the Clock app icon to access the alarm settings.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [216, 177]}<image>Thought: Click PLUS to add a new alarm.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [141, 491]}<image>",
"role": "user"
},
{
"content": "Thought: Click on the '12' to set the hour for the alarm.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [136, 271]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/6271175289399993355_6.png",
"training_data/AITW/aitw_images/general/6271175289399993355_7.png",
"training_data/AITW/aitw_images/general/6271175289399993355_8.png"
] | 2,057
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Set an alarm for 12pm<image>Thought: Click PLUS to add a new alarm.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [141, 491]}<image>Thought: Click on the '12' to set the hour for the alarm.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [136, 271]}<image>",
"role": "user"
},
{
"content": "Thought: Select PM to ensure the alarm is set for 12 in the afternoon.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [202, 221]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/6271175289399993355_7.png",
"training_data/AITW/aitw_images/general/6271175289399993355_8.png",
"training_data/AITW/aitw_images/general/6271175289399993355_9.png"
] | 2,058
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Set an alarm for 12pm<image>Thought: Click on the '12' to set the hour for the alarm.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [136, 271]}<image>Thought: Select PM to ensure the alarm is set for 12 in the afternoon.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [202, 221]}<image>",
"role": "user"
},
{
"content": "Thought: Select the time by tapping on \"12\" to set the hour for the alarm.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [210, 445]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/6271175289399993355_8.png",
"training_data/AITW/aitw_images/general/6271175289399993355_9.png",
"training_data/AITW/aitw_images/general/6271175289399993355_10.png"
] | 2,059
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Set an alarm for 12pm<image>Thought: Select PM to ensure the alarm is set for 12 in the afternoon.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [202, 221]}<image>Thought: Select the time by tapping on \"12\" to set the hour for the alarm.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [210, 445]}<image>",
"role": "user"
},
{
"content": "Thought: You successfully set the alarm by interacting with the time display for 12:00 PM in the Alarm GUI.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/6271175289399993355_9.png",
"training_data/AITW/aitw_images/general/6271175289399993355_10.png",
"training_data/AITW/aitw_images/general/6271175289399993355_11.png"
] | 2,060
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Toggle notifications for the Google Photos app<image>",
"role": "user"
},
{
"content": "Thought: Press home to return to the main screen where app settings can be accessed.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4199279418584539208_0.png"
] | 2,061
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Toggle notifications for the Google Photos app<image>Thought: Press home to return to the main screen where app settings can be accessed.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Scroll down to access the settings menu where notification options are located for Google Photos.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4199279418584539208_0.png",
"training_data/AITW/aitw_images/general/4199279418584539208_1.png"
] | 2,062
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Toggle notifications for the Google Photos app<image>Thought: Press home to return to the main screen where app settings can be accessed.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>Thought: Scroll down to access the settings menu where notification options are located for Google Photos.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Click the Photos app icon to open Google Photos and access its settings.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [634, 919]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4199279418584539208_0.png",
"training_data/AITW/aitw_images/general/4199279418584539208_1.png",
"training_data/AITW/aitw_images/general/4199279418584539208_2.png"
] | 2,063
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Toggle notifications for the Google Photos app<image>Thought: Scroll down to access the settings menu where notification options are located for Google Photos.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>Thought: Click the Photos app icon to open Google Photos and access its settings.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [634, 919]}<image>",
"role": "user"
},
{
"content": "Thought: Click the \"Update now\" button to initiate the update process for Google Photos. Since the current version is unavailable, this action is necessary to use the app. Updating may enable new features, including notification options for Google Photos.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [588, 778]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4199279418584539208_1.png",
"training_data/AITW/aitw_images/general/4199279418584539208_2.png",
"training_data/AITW/aitw_images/general/4199279418584539208_3.png"
] | 2,064
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Toggle notifications for the Google Photos app<image>Thought: Click the Photos app icon to open Google Photos and access its settings.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [634, 919]}<image>Thought: Click the \"Update now\" button to initiate the update process for Google Photos. Since the current version is unavailable, this action is necessary to use the app. Updating may enable new features, including notification options for Google Photos.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [588, 778]}<image>",
"role": "user"
},
{
"content": "Thought: Click the Update button to ensure the Google Photos app has the latest features, which may include notification settings.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [184, 337]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4199279418584539208_2.png",
"training_data/AITW/aitw_images/general/4199279418584539208_3.png",
"training_data/AITW/aitw_images/general/4199279418584539208_4.png"
] | 2,065
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Toggle notifications for the Google Photos app<image>Thought: Click the \"Update now\" button to initiate the update process for Google Photos. Since the current version is unavailable, this action is necessary to use the app. Updating may enable new features, including notification options for Google Photos.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [588, 778]}<image>Thought: Click the Update button to ensure the Google Photos app has the latest features, which may include notification settings.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [184, 337]}<image>",
"role": "user"
},
{
"content": "Thought: Click Open on the Google Photos app to access its settings for notifications.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [507, 356]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4199279418584539208_3.png",
"training_data/AITW/aitw_images/general/4199279418584539208_4.png",
"training_data/AITW/aitw_images/general/4199279418584539208_5.png"
] | 2,066
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Toggle notifications for the Google Photos app<image>Thought: Click the Update button to ensure the Google Photos app has the latest features, which may include notification settings.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [184, 337]}<image>Thought: Click Open on the Google Photos app to access its settings for notifications.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [507, 356]}<image>",
"role": "user"
},
{
"content": "Thought: Click the \"Turn on Backup\" button to enable notifications for Google Photos and start backing up.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [521, 1195]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4199279418584539208_4.png",
"training_data/AITW/aitw_images/general/4199279418584539208_5.png",
"training_data/AITW/aitw_images/general/4199279418584539208_6.png"
] | 2,067
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Toggle notifications for the Google Photos app<image>Thought: Click Open on the Google Photos app to access its settings for notifications.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [507, 356]}<image>Thought: Click the \"Turn on Backup\" button to enable notifications for Google Photos and start backing up.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [521, 1195]}<image>",
"role": "user"
},
{
"content": "Thought: You would click Confirm to save your choice of backup options for Google Photos notifications.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [338, 1237]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4199279418584539208_5.png",
"training_data/AITW/aitw_images/general/4199279418584539208_6.png",
"training_data/AITW/aitw_images/general/4199279418584539208_7.png"
] | 2,068
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Toggle notifications for the Google Photos app<image>Thought: Click the \"Turn on Backup\" button to enable notifications for Google Photos and start backing up.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [521, 1195]}<image>Thought: You would click Confirm to save your choice of backup options for Google Photos notifications.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [338, 1237]}<image>",
"role": "user"
},
{
"content": "Thought: Click the 'Skip' button to bypass initial setup and access main Google Photos interface for notification settings.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [110, 1305]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4199279418584539208_6.png",
"training_data/AITW/aitw_images/general/4199279418584539208_7.png",
"training_data/AITW/aitw_images/general/4199279418584539208_8.png"
] | 2,069
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Toggle notifications for the Google Photos app<image>Thought: You would click Confirm to save your choice of backup options for Google Photos notifications.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [338, 1237]}<image>Thought: Click the 'Skip' button to bypass initial setup and access main Google Photos interface for notification settings.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [110, 1305]}<image>",
"role": "user"
},
{
"content": "Thought: Click Google Photos to access the app settings for toggling notifications.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [661, 102]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4199279418584539208_7.png",
"training_data/AITW/aitw_images/general/4199279418584539208_8.png",
"training_data/AITW/aitw_images/general/4199279418584539208_9.png"
] | 2,070
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Toggle notifications for the Google Photos app<image>Thought: Click the 'Skip' button to bypass initial setup and access main Google Photos interface for notification settings.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [110, 1305]}<image>Thought: Click Google Photos to access the app settings for toggling notifications.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [661, 102]}<image>",
"role": "user"
},
{
"content": "Thought: Click Photos settings to access the notification options for Google Photos.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [201, 1008]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4199279418584539208_8.png",
"training_data/AITW/aitw_images/general/4199279418584539208_9.png",
"training_data/AITW/aitw_images/general/4199279418584539208_10.png"
] | 2,071
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Toggle notifications for the Google Photos app<image>Thought: Click Google Photos to access the app settings for toggling notifications.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [661, 102]}<image>Thought: Click Photos settings to access the notification options for Google Photos.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [201, 1008]}<image>",
"role": "user"
},
{
"content": "Thought: Click on the Notifications option to access settings for app notifications.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [140, 485]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4199279418584539208_9.png",
"training_data/AITW/aitw_images/general/4199279418584539208_10.png",
"training_data/AITW/aitw_images/general/4199279418584539208_11.png"
] | 2,072
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Toggle notifications for the Google Photos app<image>Thought: Click Photos settings to access the notification options for Google Photos.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [201, 1008]}<image>Thought: Click on the Notifications option to access settings for app notifications.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [140, 485]}<image>",
"role": "user"
},
{
"content": "Thought: Click the \"Notification settings on this device\" option to access further notification management for Google Photos.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [278, 1192]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4199279418584539208_10.png",
"training_data/AITW/aitw_images/general/4199279418584539208_11.png",
"training_data/AITW/aitw_images/general/4199279418584539208_12.png"
] | 2,073
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Toggle notifications for the Google Photos app<image>Thought: Click on the Notifications option to access settings for app notifications.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [140, 485]}<image>Thought: Click the \"Notification settings on this device\" option to access further notification management for Google Photos.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [278, 1192]}<image>",
"role": "user"
},
{
"content": "Thought: Click the \"All Photos notifications\" toggle to modify notification settings for Google Photos.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [636, 397]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4199279418584539208_11.png",
"training_data/AITW/aitw_images/general/4199279418584539208_12.png",
"training_data/AITW/aitw_images/general/4199279418584539208_13.png"
] | 2,074
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Toggle notifications for the Google Photos app<image>Thought: Click the \"Notification settings on this device\" option to access further notification management for Google Photos.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [278, 1192]}<image>Thought: Click the \"All Photos notifications\" toggle to modify notification settings for Google Photos.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [636, 397]}<image>",
"role": "user"
},
{
"content": "Thought: Toggle the switch for \"All Photos notifications\" to enable or disable notifications. Since it shows status task complete, confirm that notifications have been successfully toggled.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/4199279418584539208_12.png",
"training_data/AITW/aitw_images/general/4199279418584539208_13.png",
"training_data/AITW/aitw_images/general/4199279418584539208_14.png"
] | 2,075
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Open a new window in Chrome<image>",
"role": "user"
},
{
"content": "Thought: Click the Chrome icon to open the app and create a new window.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [438, 1199]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/2845494594604571859_0.png"
] | 2,076
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Open a new window in Chrome<image>Thought: Click the Chrome icon to open the app and create a new window.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [438, 1199]}<image>",
"role": "user"
},
{
"content": "Thought: Click the \"New tab\" option to open a new window in Chrome.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [473, 1204]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/2845494594604571859_0.png",
"training_data/AITW/aitw_images/general/2845494594604571859_1.png"
] | 2,077
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Open a new window in Chrome<image>Thought: Click the Chrome icon to open the app and create a new window.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [438, 1199]}<image>Thought: Click the \"New tab\" option to open a new window in Chrome.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [473, 1204]}<image>",
"role": "user"
},
{
"content": "Thought: Click on the Chrome app to launch it and access new tab options.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [435, 1224]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/2845494594604571859_0.png",
"training_data/AITW/aitw_images/general/2845494594604571859_1.png",
"training_data/AITW/aitw_images/general/2845494594604571859_2.png"
] | 2,078
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Open a new window in Chrome<image>Thought: Click the \"New tab\" option to open a new window in Chrome.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [473, 1204]}<image>Thought: Click on the Chrome app to launch it and access new tab options.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [435, 1224]}<image>",
"role": "user"
},
{
"content": "Thought: Click the Chrome icon to open it for a new window.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [434, 1218]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/2845494594604571859_1.png",
"training_data/AITW/aitw_images/general/2845494594604571859_2.png",
"training_data/AITW/aitw_images/general/2845494594604571859_3.png"
] | 2,079
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Open a new window in Chrome<image>Thought: Click on the Chrome app to launch it and access new tab options.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [435, 1224]}<image>Thought: Click the Chrome icon to open it for a new window.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [434, 1218]}<image>",
"role": "user"
},
{
"content": "Thought: Click the THREE DOTS to access the Chrome menu for opening a new window.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [591, 96]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/2845494594604571859_2.png",
"training_data/AITW/aitw_images/general/2845494594604571859_3.png",
"training_data/AITW/aitw_images/general/2845494594604571859_4.png"
] | 2,080
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Open a new window in Chrome<image>Thought: Click the Chrome icon to open it for a new window.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [434, 1218]}<image>Thought: Click the THREE DOTS to access the Chrome menu for opening a new window.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [591, 96]}<image>",
"role": "user"
},
{
"content": "Thought: Click the PLUS icon to open a new tab in Chrome.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [322, 274]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/2845494594604571859_3.png",
"training_data/AITW/aitw_images/general/2845494594604571859_4.png",
"training_data/AITW/aitw_images/general/2845494594604571859_5.png"
] | 2,081
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Open a new window in Chrome<image>Thought: Click the THREE DOTS to access the Chrome menu for opening a new window.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [591, 96]}<image>Thought: Click the PLUS icon to open a new tab in Chrome.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [322, 274]}<image>",
"role": "user"
},
{
"content": "Thought: You have completed the action by successfully opening a new window in Chrome.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/2845494594604571859_4.png",
"training_data/AITW/aitw_images/general/2845494594604571859_5.png",
"training_data/AITW/aitw_images/general/2845494594604571859_6.png"
] | 2,082
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What is the capital of Germany?<image>",
"role": "user"
},
{
"content": "Thought: Scroll down to access more information or apps that can help find the capital of Germany.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/2527802007953686769_0.png"
] | 2,083
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What is the capital of Germany?<image>Thought: Scroll down to access more information or apps that can help find the capital of Germany.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Click on Chrome to open the web browser for searching information.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [162, 206]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/2527802007953686769_0.png",
"training_data/AITW/aitw_images/general/2527802007953686769_1.png"
] | 2,084
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What is the capital of Germany?<image>Thought: Scroll down to access more information or apps that can help find the capital of Germany.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>Thought: Click on Chrome to open the web browser for searching information.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [162, 206]}<image>",
"role": "user"
},
{
"content": "Thought: Click the Pizzahut.ca link to access the website for information.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [102, 43]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/2527802007953686769_0.png",
"training_data/AITW/aitw_images/general/2527802007953686769_1.png",
"training_data/AITW/aitw_images/general/2527802007953686769_2.png"
] | 2,085
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What is the capital of Germany?<image>Thought: Click on Chrome to open the web browser for searching information.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [162, 206]}<image>Thought: Click the Pizzahut.ca link to access the website for information.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [102, 43]}<image>",
"role": "user"
},
{
"content": "Thought: Type \"capital of Germany\" in the search bar to find the answer.\nAction: {\"action\": \"TYPE\", \"value\": \"capital of Germany\", \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/2527802007953686769_1.png",
"training_data/AITW/aitw_images/general/2527802007953686769_2.png",
"training_data/AITW/aitw_images/general/2527802007953686769_3.png"
] | 2,086
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What is the capital of Germany?<image>Thought: Click the Pizzahut.ca link to access the website for information.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [102, 43]}<image>Thought: Type \"capital of Germany\" in the search bar to find the answer.\nAction: {\"action\": \"TYPE\", \"value\": \"capital of Germany\", \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Press enter to submit the search query in the search bar.\nAction: {\"action\": \"PRESS ENTER\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/2527802007953686769_2.png",
"training_data/AITW/aitw_images/general/2527802007953686769_3.png",
"training_data/AITW/aitw_images/general/2527802007953686769_4.png"
] | 2,087
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: What is the capital of Germany?<image>Thought: Type \"capital of Germany\" in the search bar to find the answer.\nAction: {\"action\": \"TYPE\", \"value\": \"capital of Germany\", \"position\": null}<image>Thought: Press enter to submit the search query in the search bar.\nAction: {\"action\": \"PRESS ENTER\", \"value\": null, \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: The user searched for \"capital of Germany\" to find information directly. The result indicates Berlin as the capital clearly. The additional details below confirm Berlin's significance and history, making the task complete.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/2527802007953686769_3.png",
"training_data/AITW/aitw_images/general/2527802007953686769_4.png",
"training_data/AITW/aitw_images/general/2527802007953686769_5.png"
] | 2,088
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Show me some nice wallpapers for my computer<image>",
"role": "user"
},
{
"content": "Thought: Scroll down to view more options for wallpapers and explore new designs available.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/1229996165410296252_0.png"
] | 2,089
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Show me some nice wallpapers for my computer<image>Thought: Scroll down to view more options for wallpapers and explore new designs available.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Click on the \"Files\" app to access stored images and wallpapers for selection.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [93, 109]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/1229996165410296252_0.png",
"training_data/AITW/aitw_images/general/1229996165410296252_1.png"
] | 2,090
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Show me some nice wallpapers for my computer<image>Thought: Scroll down to view more options for wallpapers and explore new designs available.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>Thought: Click on the \"Files\" app to access stored images and wallpapers for selection.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [93, 109]}<image>",
"role": "user"
},
{
"content": "Thought: Click Images to view and select wallpapers for my computer.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [75, 107]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/1229996165410296252_0.png",
"training_data/AITW/aitw_images/general/1229996165410296252_1.png",
"training_data/AITW/aitw_images/general/1229996165410296252_2.png"
] | 2,091
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Show me some nice wallpapers for my computer<image>Thought: Click on the \"Files\" app to access stored images and wallpapers for selection.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [93, 109]}<image>Thought: Click Images to view and select wallpapers for my computer.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [75, 107]}<image>",
"role": "user"
},
{
"content": "Thought: Press home to exit the current image view and access other apps for wallpapers.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/1229996165410296252_1.png",
"training_data/AITW/aitw_images/general/1229996165410296252_2.png",
"training_data/AITW/aitw_images/general/1229996165410296252_3.png"
] | 2,092
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Show me some nice wallpapers for my computer<image>Thought: Click Images to view and select wallpapers for my computer.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [75, 107]}<image>Thought: Press home to exit the current image view and access other apps for wallpapers.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Click the 'M' to open the menu where you can find wallpaper options.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [132, 284]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/1229996165410296252_2.png",
"training_data/AITW/aitw_images/general/1229996165410296252_3.png",
"training_data/AITW/aitw_images/general/1229996165410296252_4.png"
] | 2,093
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Show me some nice wallpapers for my computer<image>Thought: Press home to exit the current image view and access other apps for wallpapers.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>Thought: Click the 'M' to open the menu where you can find wallpaper options.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [132, 284]}<image>",
"role": "user"
},
{
"content": "Thought: Click on the wallpapers option to access different wallpaper choices for my computer.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [179, 239]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/1229996165410296252_3.png",
"training_data/AITW/aitw_images/general/1229996165410296252_4.png",
"training_data/AITW/aitw_images/general/1229996165410296252_5.png"
] | 2,094
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Show me some nice wallpapers for my computer<image>Thought: Click the 'M' to open the menu where you can find wallpaper options.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [132, 284]}<image>Thought: Click on the wallpapers option to access different wallpaper choices for my computer.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [179, 239]}<image>",
"role": "user"
},
{
"content": "Thought: You have successfully found and displayed various wallpapers in the \"Live Wallpapers\" section.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/1229996165410296252_4.png",
"training_data/AITW/aitw_images/general/1229996165410296252_5.png",
"training_data/AITW/aitw_images/general/1229996165410296252_6.png"
] | 2,095
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Google the capital of Panama<image>",
"role": "user"
},
{
"content": "Thought: Press home to exit the email setup screen and access the home screen for further actions.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/9712982572866919591_0.png"
] | 2,096
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Google the capital of Panama<image>Thought: Press home to exit the email setup screen and access the home screen for further actions.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Scroll down to see the Chrome app on the home screen to initiate a web search.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/9712982572866919591_0.png",
"training_data/AITW/aitw_images/general/9712982572866919591_1.png"
] | 2,097
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Google the capital of Panama<image>Thought: Press home to exit the email setup screen and access the home screen for further actions.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>Thought: Scroll down to see the Chrome app on the home screen to initiate a web search.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>",
"role": "user"
},
{
"content": "Thought: Click the Chrome icon to open the web browser for searching information.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [271, 436]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/9712982572866919591_0.png",
"training_data/AITW/aitw_images/general/9712982572866919591_1.png",
"training_data/AITW/aitw_images/general/9712982572866919591_2.png"
] | 2,098
|
[
{
"content": "You are an assistant trained to navigate the mobile phone. \nGiven a task instruction, a screen observation, and an action history sequence, \noutput the next action and wait for the next observation. \nHere is the action space:\n1. `CLICK`: Click on an element, value is not applicable and the position [x,y] is required. \n2. `TYPE`: Type a string into an element, value is a string to type and the position is not applicable.\n3. `SELECT`: Select a value for an element, value is the value to select and the position is not applicable.\n4. `SCROLL UP`: Scroll up for the screen.\n5. `SCROLL DOWN`: Scroll down for the screen.\n6. `SCROLL LEFT`: Scroll left for the screen.\n7. `SCROLL RIGHT`: Scroll right for the screen.\n8. `PRESS BACK`: Press for returning to the previous step, value and position are not applicable.\n9. `PRESS HOME`: Press for returning to the home screen, value and position are not applicable.\n10. `PRESS ENTER`: Press for submitting the input content, value and position are not applicable.\n11. `STATUS TASK COMPLETE`: Indicate the task is completed, value and position are not applicable.\n12. `STATUS TASK IMPOSSIBLE `: Indicate the task is impossible to complete, value and position are not applicable.\n\nFormat your response as\nThought: <your reasoning process>\nAction: <the next action>\n\nFormat the action as a JSON object with the following keys:\n{\"action\": \"ACTION_TYPE\", \"value\": \"element\", \"position\": [x,y]}\n\nIf value or position is not applicable, set it as `None`.\nPosition represents the absolute pixel coordinates on the screenshot.\nTask: Google the capital of Panama<image>Thought: Scroll down to see the Chrome app on the home screen to initiate a web search.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>Thought: Click the Chrome icon to open the web browser for searching information.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [271, 436]}<image>",
"role": "user"
},
{
"content": "Thought: Click the search bar to begin typing the query about Panama's capital.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [227, 69]}",
"role": "assistant"
}
] |
[
"training_data/AITW/aitw_images/general/9712982572866919591_1.png",
"training_data/AITW/aitw_images/general/9712982572866919591_2.png",
"training_data/AITW/aitw_images/general/9712982572866919591_3.png"
] | 2,099
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.