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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What's the news in South Korea?<image>Thought: Click the THREE DOTS to access more options related to the news link about South Korea.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.85, 0.14]}<image>Thought: Click on the \"South Korea | Today's latest from Al Jazeera\" to access updated news specific to South Korea.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.52, 0.52]}<image>", "role": "user" }, { "content": "Thought: The user seems to have found relevant news about South Korea. Completing the task suggests satisfaction with the information displayed.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/15437655504158179263_11.png", "training_data/AITW/aitw_images/general/15437655504158179263_12.png", "training_data/AITW/aitw_images/general/15437655504158179263_13.png" ]
300
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Toggle notifications for the Google Maps app<image>", "role": "user" }, { "content": "Thought: Scroll down to view necessary options for managing app notifications.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/1884054731435794316_0.png" ]
301
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Toggle notifications for the Google Maps app<image>Thought: Scroll down to view necessary options for managing app notifications.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>", "role": "user" }, { "content": "Thought: Click the Maps icon to access the app settings for notifications.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.86, 0.5]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/1884054731435794316_0.png", "training_data/AITW/aitw_images/general/1884054731435794316_1.png" ]
302
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Toggle notifications for the Google Maps app<image>Thought: Scroll down to view necessary options for managing app notifications.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>Thought: Click the Maps icon to access the app settings for notifications.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.86, 0.5]}<image>", "role": "user" }, { "content": "Thought: Click the X to close the notification settings for Google Maps.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.91, 0.9]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/1884054731435794316_0.png", "training_data/AITW/aitw_images/general/1884054731435794316_1.png", "training_data/AITW/aitw_images/general/1884054731435794316_2.png" ]
303
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Toggle notifications for the Google Maps app<image>Thought: Click the Maps icon to access the app settings for notifications.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.86, 0.5]}<image>Thought: Click the X to close the notification settings for Google Maps.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.91, 0.9]}<image>", "role": "user" }, { "content": "Thought: Click the MIC icon to access voice commands for toggling notifications in the Google Maps app.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.9, 0.07]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/1884054731435794316_1.png", "training_data/AITW/aitw_images/general/1884054731435794316_2.png", "training_data/AITW/aitw_images/general/1884054731435794316_3.png" ]
304
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Toggle notifications for the Google Maps app<image>Thought: Click the X to close the notification settings for Google Maps.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.91, 0.9]}<image>Thought: Click the MIC icon to access voice commands for toggling notifications in the Google Maps app.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.9, 0.07]}<image>", "role": "user" }, { "content": "Thought: Click the X to close the Google account menu and return to the main screen where notifications can be managed.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.09, 0.15]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/1884054731435794316_2.png", "training_data/AITW/aitw_images/general/1884054731435794316_3.png", "training_data/AITW/aitw_images/general/1884054731435794316_4.png" ]
305
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Toggle notifications for the Google Maps app<image>Thought: Click the MIC icon to access voice commands for toggling notifications in the Google Maps app.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.9, 0.07]}<image>Thought: Click the X to close the Google account menu and return to the main screen where notifications can be managed.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.09, 0.15]}<image>", "role": "user" }, { "content": "Thought: Access the notification settings for the Google Maps app. Toggle notifications to receive updates or alerts. Confirm that the notifications are successfully enabled or disabled.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/1884054731435794316_3.png", "training_data/AITW/aitw_images/general/1884054731435794316_4.png", "training_data/AITW/aitw_images/general/1884054731435794316_5.png" ]
306
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Check the settings for the Google Play Music app<image>", "role": "user" }, { "content": "Thought: Press home to return to the main screen where apps are accessible for checking Google Play Music settings.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/14737808430622259279_0.png" ]
307
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Check the settings for the Google Play Music app<image>Thought: Press home to return to the main screen where apps are accessible for checking Google Play Music settings.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>", "role": "user" }, { "content": "Thought: Tap the settings icon to open app settings and check Google Play Music configuration.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/14737808430622259279_0.png", "training_data/AITW/aitw_images/general/14737808430622259279_1.png" ]
308
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Checkthe settings for the Amazon Prime Music app<image>", "role": "user" }, { "content": "Thought: Swipe through the app icons to find the Amazon Prime Music app. After confirming it is not displayed, conclude that accessing its settings is impossible from this screen.\nAction: {\"action\": \"STATUS TASK IMPOSSIBLE\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/3057497723905774014_0.png" ]
309
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from Mexico city to Zurich<image>", "role": "user" }, { "content": "Thought: Scroll down to access more options or features related to flight searches on the home screen.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/119559407882849615_0.png" ]
310
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from Mexico city to Zurich<image>Thought: Scroll down to access more options or features related to flight searches on the home screen.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>", "role": "user" }, { "content": "Thought: Click GOOGLE to access the search engine for flight information.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.69, 0.48]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/119559407882849615_0.png", "training_data/AITW/aitw_images/general/119559407882849615_1.png" ]
311
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from Mexico city to Zurich<image>Thought: Scroll down to access more options or features related to flight searches on the home screen.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>Thought: Click GOOGLE to access the search engine for flight information.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.69, 0.48]}<image>", "role": "user" }, { "content": "Thought: Click on \"Singapore\" to gather information that may help in planning travels or flights.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.7, 0.14]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/119559407882849615_0.png", "training_data/AITW/aitw_images/general/119559407882849615_1.png", "training_data/AITW/aitw_images/general/119559407882849615_2.png" ]
312
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from Mexico city to Zurich<image>Thought: Click GOOGLE to access the search engine for flight information.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.69, 0.48]}<image>Thought: Click on \"Singapore\" to gather information that may help in planning travels or flights.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.7, 0.14]}<image>", "role": "user" }, { "content": "Thought: Click the 'X' to clear the current search input and start a new search for flights.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.9, 0.07]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/119559407882849615_1.png", "training_data/AITW/aitw_images/general/119559407882849615_2.png", "training_data/AITW/aitw_images/general/119559407882849615_3.png" ]
313
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from Mexico city to Zurich<image>Thought: Click on \"Singapore\" to gather information that may help in planning travels or flights.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.7, 0.14]}<image>Thought: Click the 'X' to clear the current search input and start a new search for flights.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.9, 0.07]}<image>", "role": "user" }, { "content": "Thought: Click the search button to execute the flight search from Mexico City to Zurich.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.28, 0.06]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/119559407882849615_2.png", "training_data/AITW/aitw_images/general/119559407882849615_3.png", "training_data/AITW/aitw_images/general/119559407882849615_4.png" ]
314
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from Mexico city to Zurich<image>Thought: Click the 'X' to clear the current search input and start a new search for flights.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.9, 0.07]}<image>Thought: Click the search button to execute the flight search from Mexico City to Zurich.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.28, 0.06]}<image>", "role": "user" }, { "content": "Thought: Type in the search bar to find flights for the specified route. This is necessary to initiate the search process.\nAction: {\"action\": \"TYPE\", \"value\": \"Search for flights from Mexico city to Zurich \", \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/119559407882849615_3.png", "training_data/AITW/aitw_images/general/119559407882849615_4.png", "training_data/AITW/aitw_images/general/119559407882849615_5.png" ]
315
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from Mexico city to Zurich<image>Thought: Click the search button to execute the flight search from Mexico City to Zurich.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.28, 0.06]}<image>Thought: Type in the search bar to find flights for the specified route. This is necessary to initiate the search process.\nAction: {\"action\": \"TYPE\", \"value\": \"Search for flights from Mexico city to Zurich \", \"position\": null}<image>", "role": "user" }, { "content": "Thought: Click on Zurich in the search results to select the desired destination for flights.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.33, 0.14]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/119559407882849615_4.png", "training_data/AITW/aitw_images/general/119559407882849615_5.png", "training_data/AITW/aitw_images/general/119559407882849615_6.png" ]
316
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from Mexico city to Zurich<image>Thought: Type in the search bar to find flights for the specified route. This is necessary to initiate the search process.\nAction: {\"action\": \"TYPE\", \"value\": \"Search for flights from Mexico city to Zurich \", \"position\": null}<image>Thought: Click on Zurich in the search results to select the desired destination for flights.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.33, 0.14]}<image>", "role": "user" }, { "content": "Thought: Scroll down to view more flight options available for Mexico City to Zurich.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/119559407882849615_5.png", "training_data/AITW/aitw_images/general/119559407882849615_6.png", "training_data/AITW/aitw_images/general/119559407882849615_7.png" ]
317
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from Mexico city to Zurich<image>Thought: Click on Zurich in the search results to select the desired destination for flights.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.33, 0.14]}<image>Thought: Scroll down to view more flight options available for Mexico City to Zurich.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>", "role": "user" }, { "content": "Thought: You would click KLM to access the specific flight details from Mexico City to Zurich.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.14, 0.56]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/119559407882849615_6.png", "training_data/AITW/aitw_images/general/119559407882849615_7.png", "training_data/AITW/aitw_images/general/119559407882849615_8.png" ]
318
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from Mexico city to Zurich<image>Thought: Scroll down to view more flight options available for Mexico City to Zurich.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>Thought: You would click KLM to access the specific flight details from Mexico City to Zurich.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.14, 0.56]}<image>", "role": "user" }, { "content": "Thought: The user completed the task of searching for flights, indicating they successfully navigated the website despite the cookie consent popup.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/119559407882849615_7.png", "training_data/AITW/aitw_images/general/119559407882849615_8.png", "training_data/AITW/aitw_images/general/119559407882849615_9.png" ]
319
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What's the news in Myanmar?<image>", "role": "user" }, { "content": "Thought: Press home to exit the email setup screen and return to the main interface for easier news access.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/3882928029571941012_0.png" ]
320
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What's the news in Myanmar?<image>Thought: Press home to exit the email setup screen and return to the main interface for easier news access.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>", "role": "user" }, { "content": "Thought: Click the Google app to access the search engine for news updates on Myanmar.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.3, 0.63]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/3882928029571941012_0.png", "training_data/AITW/aitw_images/general/3882928029571941012_1.png" ]
321
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What's the news in Myanmar?<image>Thought: Press home to exit the email setup screen and return to the main interface for easier news access.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>Thought: Click the Google app to access the search engine for news updates on Myanmar.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.3, 0.63]}<image>", "role": "user" }, { "content": "Thought: Click GOOGLE to access the search engine for current news in Myanmar.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.33, 0.64]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/3882928029571941012_0.png", "training_data/AITW/aitw_images/general/3882928029571941012_1.png", "training_data/AITW/aitw_images/general/3882928029571941012_2.png" ]
322
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What's the news in Myanmar?<image>Thought: Click the Google app to access the search engine for news updates on Myanmar.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.3, 0.63]}<image>Thought: Click GOOGLE to access the search engine for current news in Myanmar.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.33, 0.64]}<image>", "role": "user" }, { "content": "Thought: Click Google to access the internet for news about Myanmar.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.28, 0.64]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/3882928029571941012_1.png", "training_data/AITW/aitw_images/general/3882928029571941012_2.png", "training_data/AITW/aitw_images/general/3882928029571941012_3.png" ]
323
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What's the news in Myanmar?<image>Thought: Click GOOGLE to access the search engine for current news in Myanmar.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.33, 0.64]}<image>Thought: Click Google to access the internet for news about Myanmar.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.28, 0.64]}<image>", "role": "user" }, { "content": "Thought: Click Google to access the search engine for news updates.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.33, 0.63]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/3882928029571941012_2.png", "training_data/AITW/aitw_images/general/3882928029571941012_3.png", "training_data/AITW/aitw_images/general/3882928029571941012_4.png" ]
324
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What's the news in Myanmar?<image>Thought: Click Google to access the internet for news about Myanmar.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.28, 0.64]}<image>Thought: Click Google to access the search engine for news updates.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.33, 0.63]}<image>", "role": "user" }, { "content": "Thought: Click GOOGLE to access the internet and search for news related to Myanmar.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.33, 0.63]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/3882928029571941012_3.png", "training_data/AITW/aitw_images/general/3882928029571941012_4.png", "training_data/AITW/aitw_images/general/3882928029571941012_5.png" ]
325
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What's the news in Myanmar?<image>Thought: Click Google to access the search engine for news updates.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.33, 0.63]}<image>Thought: Click GOOGLE to access the internet and search for news related to Myanmar.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.33, 0.63]}<image>", "role": "user" }, { "content": "Thought: Click the search bar to enter a query about news in Myanmar instead of flight information.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.42, 0.22]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/3882928029571941012_4.png", "training_data/AITW/aitw_images/general/3882928029571941012_5.png", "training_data/AITW/aitw_images/general/3882928029571941012_6.png" ]
326
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What's the news in Myanmar?<image>Thought: Click GOOGLE to access the internet and search for news related to Myanmar.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.33, 0.63]}<image>Thought: Click the search bar to enter a query about news in Myanmar instead of flight information.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.42, 0.22]}<image>", "role": "user" }, { "content": "Thought: Click the X to clear the current search query regarding flights and start a new search for news in Myanmar.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.9, 0.06]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/3882928029571941012_5.png", "training_data/AITW/aitw_images/general/3882928029571941012_6.png", "training_data/AITW/aitw_images/general/3882928029571941012_7.png" ]
327
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What's the news in Myanmar?<image>Thought: Click the search bar to enter a query about news in Myanmar instead of flight information.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.42, 0.22]}<image>Thought: Click the X to clear the current search query regarding flights and start a new search for news in Myanmar.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.9, 0.06]}<image>", "role": "user" }, { "content": "Thought: You would click on \"designer collection\" in \"trending searches\" to find relevant information, possibly about events or news.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.59, 0.07]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/3882928029571941012_6.png", "training_data/AITW/aitw_images/general/3882928029571941012_7.png", "training_data/AITW/aitw_images/general/3882928029571941012_8.png" ]
328
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What's the news in Myanmar?<image>Thought: Click the X to clear the current search query regarding flights and start a new search for news in Myanmar.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.9, 0.06]}<image>Thought: You would click on \"designer collection\" in \"trending searches\" to find relevant information, possibly about events or news.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.59, 0.07]}<image>", "role": "user" }, { "content": "Thought: Type \"What's the news in Myanmar?\" in the search bar to find relevant updates.\nAction: {\"action\": \"TYPE\", \"value\": \"What's the news in Myanmar? \", \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/3882928029571941012_7.png", "training_data/AITW/aitw_images/general/3882928029571941012_8.png", "training_data/AITW/aitw_images/general/3882928029571941012_9.png" ]
329
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What's the news in Myanmar?<image>Thought: You would click on \"designer collection\" in \"trending searches\" to find relevant information, possibly about events or news.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.59, 0.07]}<image>Thought: Type \"What's the news in Myanmar?\" in the search bar to find relevant updates.\nAction: {\"action\": \"TYPE\", \"value\": \"What's the news in Myanmar? \", \"position\": null}<image>", "role": "user" }, { "content": "Thought: Click on \"what's the news in myanmar\" to initiate a search for current news articles.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.37, 0.16]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/3882928029571941012_8.png", "training_data/AITW/aitw_images/general/3882928029571941012_9.png", "training_data/AITW/aitw_images/general/3882928029571941012_10.png" ]
330
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What's the news in Myanmar?<image>Thought: Type \"What's the news in Myanmar?\" in the search bar to find relevant updates.\nAction: {\"action\": \"TYPE\", \"value\": \"What's the news in Myanmar? \", \"position\": null}<image>Thought: Click on \"what's the news in myanmar\" to initiate a search for current news articles.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.37, 0.16]}<image>", "role": "user" }, { "content": "Thought: Scroll down to view more news articles and updates about Myanmar.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/3882928029571941012_9.png", "training_data/AITW/aitw_images/general/3882928029571941012_10.png", "training_data/AITW/aitw_images/general/3882928029571941012_11.png" ]
331
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What's the news in Myanmar?<image>Thought: Click on \"what's the news in myanmar\" to initiate a search for current news articles.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.37, 0.16]}<image>Thought: Scroll down to view more news articles and updates about Myanmar.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>", "role": "user" }, { "content": "Thought: Click on the comment section to engage with readers or share thoughts on Myanmar news.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.34, 0.46]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/3882928029571941012_10.png", "training_data/AITW/aitw_images/general/3882928029571941012_11.png", "training_data/AITW/aitw_images/general/3882928029571941012_12.png" ]
332
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What's the news in Myanmar?<image>Thought: Scroll down to view more news articles and updates about Myanmar.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>Thought: Click on the comment section to engage with readers or share thoughts on Myanmar news.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.34, 0.46]}<image>", "role": "user" }, { "content": "Thought: Click AGREE to consent to the privacy options and proceed with accessing news content about Myanmar.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.56, 0.86]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/3882928029571941012_11.png", "training_data/AITW/aitw_images/general/3882928029571941012_12.png", "training_data/AITW/aitw_images/general/3882928029571941012_13.png" ]
333
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What's the news in Myanmar?<image>Thought: Click on the comment section to engage with readers or share thoughts on Myanmar news.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.34, 0.46]}<image>Thought: Click AGREE to consent to the privacy options and proceed with accessing news content about Myanmar.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.56, 0.86]}<image>", "role": "user" }, { "content": "Thought: Check the news section labeled 'Myanmar' for updates on recent events. The task is complete as information was gathered successfully from the page.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/3882928029571941012_12.png", "training_data/AITW/aitw_images/general/3882928029571941012_13.png", "training_data/AITW/aitw_images/general/3882928029571941012_14.png" ]
334
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Who is the president of the United States?<image>", "role": "user" }, { "content": "Thought: Press home to exit the current app and return to the main screen for further searching.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/11868572470103239004_0.png" ]
335
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Who is the president of the United States?<image>Thought: Press home to exit the current app and return to the main screen for further searching.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>", "role": "user" }, { "content": "Thought: Click the Google app to access a search engine for information about the president of the United States.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.69, 0.78]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/11868572470103239004_0.png", "training_data/AITW/aitw_images/general/11868572470103239004_1.png" ]
336
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Who is the president of the United States?<image>Thought: Press home to exit the current app and return to the main screen for further searching.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>Thought: Click the Google app to access a search engine for information about the president of the United States.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.69, 0.78]}<image>", "role": "user" }, { "content": "Thought: The user likely clicked the Google search bar to find information about the U.S. president.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.54, 0.08]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/11868572470103239004_0.png", "training_data/AITW/aitw_images/general/11868572470103239004_1.png", "training_data/AITW/aitw_images/general/11868572470103239004_2.png" ]
337
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Who is the president of the United States?<image>Thought: Click the Google app to access a search engine for information about the president of the United States.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.69, 0.78]}<image>Thought: The user likely clicked the Google search bar to find information about the U.S. president.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.54, 0.08]}<image>", "role": "user" }, { "content": "Thought: Type in the search bar to find information about the current president of the United States.\nAction: {\"action\": \"TYPE\", \"value\": \"Who is the president of the United States?\", \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/11868572470103239004_1.png", "training_data/AITW/aitw_images/general/11868572470103239004_2.png", "training_data/AITW/aitw_images/general/11868572470103239004_3.png" ]
338
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Who is the president of the United States?<image>Thought: The user likely clicked the Google search bar to find information about the U.S. president.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.54, 0.08]}<image>Thought: Type in the search bar to find information about the current president of the United States.\nAction: {\"action\": \"TYPE\", \"value\": \"Who is the president of the United States?\", \"position\": null}<image>", "role": "user" }, { "content": "Thought: Click on \"president of the\" to select that part of the search query for potential editing or submission.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.57, 0.14]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/11868572470103239004_2.png", "training_data/AITW/aitw_images/general/11868572470103239004_3.png", "training_data/AITW/aitw_images/general/11868572470103239004_4.png" ]
339
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Who is the president of the United States?<image>Thought: Type in the search bar to find information about the current president of the United States.\nAction: {\"action\": \"TYPE\", \"value\": \"Who is the president of the United States?\", \"position\": null}<image>Thought: Click on \"president of the\" to select that part of the search query for potential editing or submission.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.57, 0.14]}<image>", "role": "user" }, { "content": "Thought: Scroll down to find the answer regarding who is the president of the United States.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/11868572470103239004_3.png", "training_data/AITW/aitw_images/general/11868572470103239004_4.png", "training_data/AITW/aitw_images/general/11868572470103239004_5.png" ]
340
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Who is the president of the United States?<image>Thought: Click on \"president of the\" to select that part of the search query for potential editing or submission.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.57, 0.14]}<image>Thought: Scroll down to find the answer regarding who is the president of the United States.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>", "role": "user" }, { "content": "Thought: Click \"Get 50%\" to access relevant information about U.S. presidents on the site.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.95, 0.24]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/11868572470103239004_4.png", "training_data/AITW/aitw_images/general/11868572470103239004_5.png", "training_data/AITW/aitw_images/general/11868572470103239004_6.png" ]
341
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Who is the president of the United States?<image>Thought: Scroll down to find the answer regarding who is the president of the United States.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>Thought: Click \"Get 50%\" to access relevant information about U.S. presidents on the site.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.95, 0.24]}<image>", "role": "user" }, { "content": "Thought: The user likely found the answer by locating Joe Biden's name in the search results for U.S. President.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/11868572470103239004_5.png", "training_data/AITW/aitw_images/general/11868572470103239004_6.png", "training_data/AITW/aitw_images/general/11868572470103239004_7.png" ]
342
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What is the average speed of a car?<image>", "role": "user" }, { "content": "Thought: Click the Google Maps icon to find the average speed of a car by accessing navigation features.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.73, 0.8]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/11714679647466403324_0.png" ]
343
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What is the average speed of a car?<image>Thought: Click the Google Maps icon to find the average speed of a car by accessing navigation features.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.73, 0.8]}<image>", "role": "user" }, { "content": "Thought: Click on the search bar to input a query about average speed of a car.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.38, 0.07]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/11714679647466403324_0.png", "training_data/AITW/aitw_images/general/11714679647466403324_1.png" ]
344
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What is the average speed of a car?<image>Thought: Click the Google Maps icon to find the average speed of a car by accessing navigation features.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.73, 0.8]}<image>Thought: Click on the search bar to input a query about average speed of a car.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.38, 0.07]}<image>", "role": "user" }, { "content": "Thought: Type in the search bar to find information about average speed.\nAction: {\"action\": \"TYPE\", \"value\": \"average speed of a car\", \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/11714679647466403324_0.png", "training_data/AITW/aitw_images/general/11714679647466403324_1.png", "training_data/AITW/aitw_images/general/11714679647466403324_2.png" ]
345
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What is the average speed of a car?<image>Thought: Click on the search bar to input a query about average speed of a car.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.38, 0.07]}<image>Thought: Type in the search bar to find information about average speed.\nAction: {\"action\": \"TYPE\", \"value\": \"average speed of a car\", \"position\": null}<image>", "role": "user" }, { "content": "Thought: Click on \"average speed of a car\" in the search suggestions to view relevant information.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.49, 0.16]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/11714679647466403324_1.png", "training_data/AITW/aitw_images/general/11714679647466403324_2.png", "training_data/AITW/aitw_images/general/11714679647466403324_3.png" ]
346
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What is the average speed of a car?<image>Thought: Type in the search bar to find information about average speed.\nAction: {\"action\": \"TYPE\", \"value\": \"average speed of a car\", \"position\": null}<image>Thought: Click on \"average speed of a car\" in the search suggestions to view relevant information.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.49, 0.16]}<image>", "role": "user" }, { "content": "Thought: You completed the task after finding the information about average speed in the search results displayed.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/11714679647466403324_2.png", "training_data/AITW/aitw_images/general/11714679647466403324_3.png", "training_data/AITW/aitw_images/general/11714679647466403324_4.png" ]
347
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Google the capital of Colombia<image>", "role": "user" }, { "content": "Thought: Press home to exit the Gmail app and access the home screen for further tasks.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/7649068117262903620_0.png" ]
348
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Google the capital of Colombia<image>Thought: Press home to exit the Gmail app and access the home screen for further tasks.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>", "role": "user" }, { "content": "Thought: Search for the Google app by clicking the 'G' icon on the home screen.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.15, 0.8]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/7649068117262903620_0.png", "training_data/AITW/aitw_images/general/7649068117262903620_1.png" ]
349
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Google the capital of Colombia<image>Thought: Press home to exit the Gmail app and access the home screen for further tasks.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>Thought: Search for the Google app by clicking the 'G' icon on the home screen.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.15, 0.8]}<image>", "role": "user" }, { "content": "Thought: Type \"capital of Colombia\" in the search bar to initiate a Google search for that information.\nAction: {\"action\": \"TYPE\", \"value\": \" capital of Colombia\", \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/7649068117262903620_0.png", "training_data/AITW/aitw_images/general/7649068117262903620_1.png", "training_data/AITW/aitw_images/general/7649068117262903620_2.png" ]
350
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Google the capital of Colombia<image>Thought: Search for the Google app by clicking the 'G' icon on the home screen.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.15, 0.8]}<image>Thought: Type \"capital of Colombia\" in the search bar to initiate a Google search for that information.\nAction: {\"action\": \"TYPE\", \"value\": \" capital of Colombia\", \"position\": null}<image>", "role": "user" }, { "content": "Thought: Press enter to submit the search query for the capital of Colombia in the Google search bar.\nAction: {\"action\": \"PRESS ENTER\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/7649068117262903620_1.png", "training_data/AITW/aitw_images/general/7649068117262903620_2.png", "training_data/AITW/aitw_images/general/7649068117262903620_3.png" ]
351
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Google the capital of Colombia<image>Thought: Type \"capital of Colombia\" in the search bar to initiate a Google search for that information.\nAction: {\"action\": \"TYPE\", \"value\": \" capital of Colombia\", \"position\": null}<image>Thought: Press enter to submit the search query for the capital of Colombia in the Google search bar.\nAction: {\"action\": \"PRESS ENTER\", \"value\": null, \"position\": null}<image>", "role": "user" }, { "content": "Thought: Complete the search by verifying that \"Colombia\" appears with its capital information displayed.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/7649068117262903620_2.png", "training_data/AITW/aitw_images/general/7649068117262903620_3.png", "training_data/AITW/aitw_images/general/7649068117262903620_4.png" ]
352
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from NYC to Tokyo<image>", "role": "user" }, { "content": "Thought: Press home to return to the main screen and start a new search for flights from NYC to Tokyo.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/154322127451137180_0.png" ]
353
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from NYC to Tokyo<image>Thought: Press home to return to the main screen and start a new search for flights from NYC to Tokyo.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>", "role": "user" }, { "content": "Thought: Click the browser icon to prepare for searching flights from NYC to Tokyo.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.67, 0.8]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/154322127451137180_0.png", "training_data/AITW/aitw_images/general/154322127451137180_1.png" ]
354
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from NYC to Tokyo<image>Thought: Press home to return to the main screen and start a new search for flights from NYC to Tokyo.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>Thought: Click the browser icon to prepare for searching flights from NYC to Tokyo.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.67, 0.8]}<image>", "role": "user" }, { "content": "Thought: Click the Chrome icon to open the browser for searching flights.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.7, 0.77]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/154322127451137180_0.png", "training_data/AITW/aitw_images/general/154322127451137180_1.png", "training_data/AITW/aitw_images/general/154322127451137180_2.png" ]
355
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from NYC to Tokyo<image>Thought: Click the browser icon to prepare for searching flights from NYC to Tokyo.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.67, 0.8]}<image>Thought: Click the Chrome icon to open the browser for searching flights.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.7, 0.77]}<image>", "role": "user" }, { "content": "Thought: Click on the \"New tab\" option to open a web browser for searching flights.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.69, 0.77]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/154322127451137180_1.png", "training_data/AITW/aitw_images/general/154322127451137180_2.png", "training_data/AITW/aitw_images/general/154322127451137180_3.png" ]
356
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from NYC to Tokyo<image>Thought: Click the Chrome icon to open the browser for searching flights.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.7, 0.77]}<image>Thought: Click on the \"New tab\" option to open a web browser for searching flights.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.69, 0.77]}<image>", "role": "user" }, { "content": "Thought: Click the icon for Google Chrome to initiate a web search for flights.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.68, 0.77]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/154322127451137180_2.png", "training_data/AITW/aitw_images/general/154322127451137180_3.png", "training_data/AITW/aitw_images/general/154322127451137180_4.png" ]
357
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from NYC to Tokyo<image>Thought: Click on the \"New tab\" option to open a web browser for searching flights.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.69, 0.77]}<image>Thought: Click the icon for Google Chrome to initiate a web search for flights.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.68, 0.77]}<image>", "role": "user" }, { "content": "Thought: Click the \"New tab\" option to open a browser for flight searching.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.71, 0.78]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/154322127451137180_3.png", "training_data/AITW/aitw_images/general/154322127451137180_4.png", "training_data/AITW/aitw_images/general/154322127451137180_5.png" ]
358
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from NYC to Tokyo<image>Thought: Click the icon for Google Chrome to initiate a web search for flights.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.68, 0.77]}<image>Thought: Click the \"New tab\" option to open a browser for flight searching.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.71, 0.78]}<image>", "role": "user" }, { "content": "Thought: Click the search bar to input flight details for NYC to Tokyo. Your query suggests you're looking for information about average speed instead.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.53, 0.06]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/154322127451137180_4.png", "training_data/AITW/aitw_images/general/154322127451137180_5.png", "training_data/AITW/aitw_images/general/154322127451137180_6.png" ]
359
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from NYC to Tokyo<image>Thought: Click the \"New tab\" option to open a browser for flight searching.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.71, 0.78]}<image>Thought: Click the search bar to input flight details for NYC to Tokyo. Your query suggests you're looking for information about average speed instead.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.53, 0.06]}<image>", "role": "user" }, { "content": "Thought: Type the search query \"flights from NYC to Tokyo\" into the search bar to initiate the flight search.\nAction: {\"action\": \"TYPE\", \"value\": \"flights from NYC to Tokyo\", \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/154322127451137180_5.png", "training_data/AITW/aitw_images/general/154322127451137180_6.png", "training_data/AITW/aitw_images/general/154322127451137180_7.png" ]
360
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from NYC to Tokyo<image>Thought: Click the search bar to input flight details for NYC to Tokyo. Your query suggests you're looking for information about average speed instead.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.53, 0.06]}<image>Thought: Type the search query \"flights from NYC to Tokyo\" into the search bar to initiate the flight search.\nAction: {\"action\": \"TYPE\", \"value\": \"flights from NYC to Tokyo\", \"position\": null}<image>", "role": "user" }, { "content": "Thought: Press enter to initiate the search for flights from NYC to Tokyo using the search bar.\nAction: {\"action\": \"PRESS ENTER\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/154322127451137180_6.png", "training_data/AITW/aitw_images/general/154322127451137180_7.png", "training_data/AITW/aitw_images/general/154322127451137180_8.png" ]
361
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from NYC to Tokyo<image>Thought: Type the search query \"flights from NYC to Tokyo\" into the search bar to initiate the flight search.\nAction: {\"action\": \"TYPE\", \"value\": \"flights from NYC to Tokyo\", \"position\": null}<image>Thought: Press enter to initiate the search for flights from NYC to Tokyo using the search bar.\nAction: {\"action\": \"PRESS ENTER\", \"value\": null, \"position\": null}<image>", "role": "user" }, { "content": "Thought: Scroll down to view more flight options and details related to NYC to Tokyo.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/154322127451137180_7.png", "training_data/AITW/aitw_images/general/154322127451137180_8.png", "training_data/AITW/aitw_images/general/154322127451137180_9.png" ]
362
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from NYC to Tokyo<image>Thought: Press enter to initiate the search for flights from NYC to Tokyo using the search bar.\nAction: {\"action\": \"PRESS ENTER\", \"value\": null, \"position\": null}<image>Thought: Scroll down to view more flight options and details related to NYC to Tokyo.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>", "role": "user" }, { "content": "Thought: Scroll down to view more flight options and details relevant to traveling from NYC to Tokyo.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/154322127451137180_8.png", "training_data/AITW/aitw_images/general/154322127451137180_9.png", "training_data/AITW/aitw_images/general/154322127451137180_10.png" ]
363
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from NYC to Tokyo<image>Thought: Scroll down to view more flight options and details related to NYC to Tokyo.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>Thought: Scroll down to view more flight options and details relevant to traveling from NYC to Tokyo.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>", "role": "user" }, { "content": "Thought: Scroll down to view more flight options and details for traveling from NYC to Tokyo.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/154322127451137180_9.png", "training_data/AITW/aitw_images/general/154322127451137180_10.png", "training_data/AITW/aitw_images/general/154322127451137180_11.png" ]
364
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from NYC to Tokyo<image>Thought: Scroll down to view more flight options and details relevant to traveling from NYC to Tokyo.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>Thought: Scroll down to view more flight options and details for traveling from NYC to Tokyo.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>", "role": "user" }, { "content": "Thought: You want to view the available flight options after searching for flights. Clicking \"Show flights\" reveals the detailed flight information displayed on the screen.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.55, 0.52]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/154322127451137180_10.png", "training_data/AITW/aitw_images/general/154322127451137180_11.png", "training_data/AITW/aitw_images/general/154322127451137180_12.png" ]
365
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from NYC to Tokyo<image>Thought: Scroll down to view more flight options and details for traveling from NYC to Tokyo.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>Thought: You want to view the available flight options after searching for flights. Clicking \"Show flights\" reveals the detailed flight information displayed on the screen.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.55, 0.52]}<image>", "role": "user" }, { "content": "Thought: Click the \"Show flights\" button to view detailed options for flights from NYC to Tokyo.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.56, 0.53]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/154322127451137180_11.png", "training_data/AITW/aitw_images/general/154322127451137180_12.png", "training_data/AITW/aitw_images/general/154322127451137180_13.png" ]
366
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from NYC to Tokyo<image>Thought: You want to view the available flight options after searching for flights. Clicking \"Show flights\" reveals the detailed flight information displayed on the screen.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.55, 0.52]}<image>Thought: Click the \"Show flights\" button to view detailed options for flights from NYC to Tokyo.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.56, 0.53]}<image>", "role": "user" }, { "content": "Thought: You completed the task after successfully finding and viewing flight options from NYC to Tokyo.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/154322127451137180_12.png", "training_data/AITW/aitw_images/general/154322127451137180_13.png", "training_data/AITW/aitw_images/general/154322127451137180_14.png" ]
367
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What time is it in New York?<image>", "role": "user" }, { "content": "Thought: Scroll down to access the notification area where the time might be displayed.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/567115952537557158_0.png" ]
368
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What time is it in New York?<image>Thought: Scroll down to access the notification area where the time might be displayed.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>", "role": "user" }, { "content": "Thought: Open the app drawer by clicking on 'Google' to search for the time in New York.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.13, 0.67]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/567115952537557158_0.png", "training_data/AITW/aitw_images/general/567115952537557158_1.png" ]
369
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What time is it in New York?<image>Thought: Scroll down to access the notification area where the time might be displayed.\nAction: {\"action\": \"SCROLL DOWN\", \"value\": null, \"position\": null}<image>Thought: Open the app drawer by clicking on 'Google' to search for the time in New York.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.13, 0.67]}<image>", "role": "user" }, { "content": "Thought: Click the link for \"Reddit\" to access time-related information about New York.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.3, 0.13]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/567115952537557158_0.png", "training_data/AITW/aitw_images/general/567115952537557158_1.png", "training_data/AITW/aitw_images/general/567115952537557158_2.png" ]
370
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What time is it in New York?<image>Thought: Open the app drawer by clicking on 'Google' to search for the time in New York.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.13, 0.67]}<image>Thought: Click the link for \"Reddit\" to access time-related information about New York.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.3, 0.13]}<image>", "role": "user" }, { "content": "Thought: Click X to close the current search interface and return to the home screen for further navigation.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.91, 0.07]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/567115952537557158_1.png", "training_data/AITW/aitw_images/general/567115952537557158_2.png", "training_data/AITW/aitw_images/general/567115952537557158_3.png" ]
371
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What time is it in New York?<image>Thought: Click the link for \"Reddit\" to access time-related information about New York.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.3, 0.13]}<image>Thought: Click X to close the current search interface and return to the home screen for further navigation.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.91, 0.07]}<image>", "role": "user" }, { "content": "Thought: Type \"What time is it in New York?\" into the search bar to find the current time.\nAction: {\"action\": \"TYPE\", \"value\": \"What time is it in New York?\", \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/567115952537557158_2.png", "training_data/AITW/aitw_images/general/567115952537557158_3.png", "training_data/AITW/aitw_images/general/567115952537557158_4.png" ]
372
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What time is it in New York?<image>Thought: Click X to close the current search interface and return to the home screen for further navigation.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.91, 0.07]}<image>Thought: Type \"What time is it in New York?\" into the search bar to find the current time.\nAction: {\"action\": \"TYPE\", \"value\": \"What time is it in New York?\", \"position\": null}<image>", "role": "user" }, { "content": "Thought: Click the suggested search result to get the current time in New York City.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.38, 0.13]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/567115952537557158_3.png", "training_data/AITW/aitw_images/general/567115952537557158_4.png", "training_data/AITW/aitw_images/general/567115952537557158_5.png" ]
373
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What time is it in New York?<image>Thought: Type \"What time is it in New York?\" into the search bar to find the current time.\nAction: {\"action\": \"TYPE\", \"value\": \"What time is it in New York?\", \"position\": null}<image>Thought: Click the suggested search result to get the current time in New York City.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.38, 0.13]}<image>", "role": "user" }, { "content": "Thought: The task is complete as the current time in New York is displayed clearly at the top of the screen.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/567115952537557158_4.png", "training_data/AITW/aitw_images/general/567115952537557158_5.png", "training_data/AITW/aitw_images/general/567115952537557158_6.png" ]
374
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from Mexico city to Sydney<image>", "role": "user" }, { "content": "Thought: Click on the GOOGLE icon to access the search engine for finding flights.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.3, 0.89]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/12542843092047567094_0.png" ]
375
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from Mexico city to Sydney<image>Thought: Click on the GOOGLE icon to access the search engine for finding flights.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.3, 0.89]}<image>", "role": "user" }, { "content": "Thought: Click the search bar to start typing flight details. Type \"flights from Mexico city to Sydney\" for relevant results. Press enter to execute the search and view options.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.4, 0.07]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/12542843092047567094_0.png", "training_data/AITW/aitw_images/general/12542843092047567094_1.png" ]
376
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from Mexico city to Sydney<image>Thought: Click on the GOOGLE icon to access the search engine for finding flights.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.3, 0.89]}<image>Thought: Click the search bar to start typing flight details. Type \"flights from Mexico city to Sydney\" for relevant results. Press enter to execute the search and view options.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.4, 0.07]}<image>", "role": "user" }, { "content": "Thought: Type the search query in the search bar to find relevant flight information.\nAction: {\"action\": \"TYPE\", \"value\": \"flights from Mexico city to Sydney\", \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/12542843092047567094_0.png", "training_data/AITW/aitw_images/general/12542843092047567094_1.png", "training_data/AITW/aitw_images/general/12542843092047567094_2.png" ]
377
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from Mexico city to Sydney<image>Thought: Click the search bar to start typing flight details. Type \"flights from Mexico city to Sydney\" for relevant results. Press enter to execute the search and view options.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.4, 0.07]}<image>Thought: Type the search query in the search bar to find relevant flight information.\nAction: {\"action\": \"TYPE\", \"value\": \"flights from Mexico city to Sydney\", \"position\": null}<image>", "role": "user" }, { "content": "Thought: Click on \"mexico city to\" to finish typing the flight search query.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.4, 0.11]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/12542843092047567094_1.png", "training_data/AITW/aitw_images/general/12542843092047567094_2.png", "training_data/AITW/aitw_images/general/12542843092047567094_3.png" ]
378
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for flights from Mexico city to Sydney<image>Thought: Type the search query in the search bar to find relevant flight information.\nAction: {\"action\": \"TYPE\", \"value\": \"flights from Mexico city to Sydney\", \"position\": null}<image>Thought: Click on \"mexico city to\" to finish typing the flight search query.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.4, 0.11]}<image>", "role": "user" }, { "content": "Thought: The operation indicates that the flight search results for flights from Mexico City to Sydney have been successfully displayed.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/12542843092047567094_2.png", "training_data/AITW/aitw_images/general/12542843092047567094_3.png", "training_data/AITW/aitw_images/general/12542843092047567094_4.png" ]
379
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for hotels in New York<image>", "role": "user" }, { "content": "Thought: Click the NAV BAR CIRCLE to access the main menu for searching hotels.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.42, 0.89]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/3272203257250051465_0.png" ]
380
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for hotels in New York<image>Thought: Click the NAV BAR CIRCLE to access the main menu for searching hotels.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.42, 0.89]}<image>", "role": "user" }, { "content": "Thought: Type 'hotels in New York' into the search bar to initiate a search for accommodations in that location.\nAction: {\"action\": \"TYPE\", \"value\": \" hotels in New York\", \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/3272203257250051465_0.png", "training_data/AITW/aitw_images/general/3272203257250051465_1.png" ]
381
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for hotels in New York<image>Thought: Click the NAV BAR CIRCLE to access the main menu for searching hotels.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.42, 0.89]}<image>Thought: Type 'hotels in New York' into the search bar to initiate a search for accommodations in that location.\nAction: {\"action\": \"TYPE\", \"value\": \" hotels in New York\", \"position\": null}<image>", "role": "user" }, { "content": "Thought: Click on \"hotels in new york\" to initiate a search for available accommodations.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.3, 0.12]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/3272203257250051465_0.png", "training_data/AITW/aitw_images/general/3272203257250051465_1.png", "training_data/AITW/aitw_images/general/3272203257250051465_2.png" ]
382
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for hotels in New York<image>Thought: Type 'hotels in New York' into the search bar to initiate a search for accommodations in that location.\nAction: {\"action\": \"TYPE\", \"value\": \" hotels in New York\", \"position\": null}<image>Thought: Click on \"hotels in new york\" to initiate a search for available accommodations.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.3, 0.12]}<image>", "role": "user" }, { "content": "Thought: You clicked \"Hotels in New York\" to access specific hotel listings relevant to your search.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.31, 0.55]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/3272203257250051465_1.png", "training_data/AITW/aitw_images/general/3272203257250051465_2.png", "training_data/AITW/aitw_images/general/3272203257250051465_3.png" ]
383
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for hotels in New York<image>Thought: Click on \"hotels in new york\" to initiate a search for available accommodations.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.3, 0.12]}<image>Thought: You clicked \"Hotels in New York\" to access specific hotel listings relevant to your search.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.31, 0.55]}<image>", "role": "user" }, { "content": "Thought: Click \"Accept\" to agree to cookie preferences and proceed with the hotel search.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.65, 0.91]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/3272203257250051465_2.png", "training_data/AITW/aitw_images/general/3272203257250051465_3.png", "training_data/AITW/aitw_images/general/3272203257250051465_4.png" ]
384
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for hotels in New York<image>Thought: You clicked \"Hotels in New York\" to access specific hotel listings relevant to your search.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.31, 0.55]}<image>Thought: Click \"Accept\" to agree to cookie preferences and proceed with the hotel search.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.65, 0.91]}<image>", "role": "user" }, { "content": "Thought: Click on the Search button to initiate the hotel search in New York.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.46, 0.69]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/3272203257250051465_3.png", "training_data/AITW/aitw_images/general/3272203257250051465_4.png", "training_data/AITW/aitw_images/general/3272203257250051465_5.png" ]
385
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Search for hotels in New York<image>Thought: Click \"Accept\" to agree to cookie preferences and proceed with the hotel search.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.65, 0.91]}<image>Thought: Click on the Search button to initiate the hotel search in New York.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.46, 0.69]}<image>", "role": "user" }, { "content": "Thought: The user likely intended to search for hotels in New York and successfully found relevant listings with available options.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/3272203257250051465_4.png", "training_data/AITW/aitw_images/general/3272203257250051465_5.png", "training_data/AITW/aitw_images/general/3272203257250051465_6.png" ]
386
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What are the best selling refrigerators at Home Depot?<image>", "role": "user" }, { "content": "Thought: Click the NAV BAR CIRCLE to access the app drawer for further navigation options.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.52, 0.88]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/14231584675124793991_0.png" ]
387
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What are the best selling refrigerators at Home Depot?<image>Thought: Click the NAV BAR CIRCLE to access the app drawer for further navigation options.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.52, 0.88]}<image>", "role": "user" }, { "content": "Thought: Type the query in the search bar to find relevant information about refrigerators at Home Depot.\nAction: {\"action\": \"TYPE\", \"value\": \"What are the best selling refrigerators at Home Depot\", \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/14231584675124793991_0.png", "training_data/AITW/aitw_images/general/14231584675124793991_1.png" ]
388
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What are the best selling refrigerators at Home Depot?<image>Thought: Click the NAV BAR CIRCLE to access the app drawer for further navigation options.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.52, 0.88]}<image>Thought: Type the query in the search bar to find relevant information about refrigerators at Home Depot.\nAction: {\"action\": \"TYPE\", \"value\": \"What are the best selling refrigerators at Home Depot\", \"position\": null}<image>", "role": "user" }, { "content": "Thought: Type \"best selling refrigerators at home depot\" in the search bar to find relevant information.\nAction: {\"action\": \"TYPE\", \"value\": \" at home depot\", \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/14231584675124793991_0.png", "training_data/AITW/aitw_images/general/14231584675124793991_1.png", "training_data/AITW/aitw_images/general/14231584675124793991_2.png" ]
389
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What are the best selling refrigerators at Home Depot?<image>Thought: Type the query in the search bar to find relevant information about refrigerators at Home Depot.\nAction: {\"action\": \"TYPE\", \"value\": \"What are the best selling refrigerators at Home Depot\", \"position\": null}<image>Thought: Type \"best selling refrigerators at home depot\" in the search bar to find relevant information.\nAction: {\"action\": \"TYPE\", \"value\": \" at home depot\", \"position\": null}<image>", "role": "user" }, { "content": "Thought: Click on \"refrigerator\" to select a relevant search suggestion for best-selling refrigerators.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.49, 0.21]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/14231584675124793991_1.png", "training_data/AITW/aitw_images/general/14231584675124793991_2.png", "training_data/AITW/aitw_images/general/14231584675124793991_3.png" ]
390
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: What are the best selling refrigerators at Home Depot?<image>Thought: Type \"best selling refrigerators at home depot\" in the search bar to find relevant information.\nAction: {\"action\": \"TYPE\", \"value\": \" at home depot\", \"position\": null}<image>Thought: Click on \"refrigerator\" to select a relevant search suggestion for best-selling refrigerators.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.49, 0.21]}<image>", "role": "user" }, { "content": "Thought: You likely found the search results you needed about the best selling refrigerators at Home Depot. The information displayed indicates a successful completion of your task.\nAction: {\"action\": \"STATUS TASK COMPLETE\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/14231584675124793991_2.png", "training_data/AITW/aitw_images/general/14231584675124793991_3.png", "training_data/AITW/aitw_images/general/14231584675124793991_4.png" ]
391
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Open the downloads<image>", "role": "user" }, { "content": "Thought: Press home to return to the main screen and access apps or settings for downloads.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/14556377186823585099_0.png" ]
392
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Open the downloads<image>Thought: Press home to return to the main screen and access apps or settings for downloads.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>", "role": "user" }, { "content": "Thought: Click the LOCATION to access the area where downloads are stored.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.71, 0.78]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/14556377186823585099_0.png", "training_data/AITW/aitw_images/general/14556377186823585099_1.png" ]
393
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Open the downloads<image>Thought: Press home to return to the main screen and access apps or settings for downloads.\nAction: {\"action\": \"PRESS HOME\", \"value\": null, \"position\": null}<image>Thought: Click the LOCATION to access the area where downloads are stored.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.71, 0.78]}<image>", "role": "user" }, { "content": "Thought: Click on the URL to access the webpage, which may contain downloads.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.55, 0.06]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/14556377186823585099_0.png", "training_data/AITW/aitw_images/general/14556377186823585099_1.png", "training_data/AITW/aitw_images/general/14556377186823585099_2.png" ]
394
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Open the downloads<image>Thought: Click the LOCATION to access the area where downloads are stored.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.71, 0.78]}<image>Thought: Click on the URL to access the webpage, which may contain downloads.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.55, 0.06]}<image>", "role": "user" }, { "content": "Thought: Type \"downloads\" in the search bar to locate the downloads section.\nAction: {\"action\": \"TYPE\", \"value\": \"downloads\", \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/14556377186823585099_1.png", "training_data/AITW/aitw_images/general/14556377186823585099_2.png", "training_data/AITW/aitw_images/general/14556377186823585099_3.png" ]
395
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Open the downloads<image>Thought: Click on the URL to access the webpage, which may contain downloads.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.55, 0.06]}<image>Thought: Type \"downloads\" in the search bar to locate the downloads section.\nAction: {\"action\": \"TYPE\", \"value\": \"downloads\", \"position\": null}<image>", "role": "user" }, { "content": "Thought: Type \"download\" in the search bar to locate the Downloads section on your device.\nAction: {\"action\": \"TYPE\", \"value\": \"\", \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/14556377186823585099_2.png", "training_data/AITW/aitw_images/general/14556377186823585099_3.png", "training_data/AITW/aitw_images/general/14556377186823585099_4.png" ]
396
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Open the downloads<image>Thought: Type \"downloads\" in the search bar to locate the downloads section.\nAction: {\"action\": \"TYPE\", \"value\": \"downloads\", \"position\": null}<image>Thought: Type \"download\" in the search bar to locate the Downloads section on your device.\nAction: {\"action\": \"TYPE\", \"value\": \"\", \"position\": null}<image>", "role": "user" }, { "content": "Thought: Click the \"X\" to remove the search term and access the main menu options for downloads.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.92, 0.06]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/14556377186823585099_3.png", "training_data/AITW/aitw_images/general/14556377186823585099_4.png", "training_data/AITW/aitw_images/general/14556377186823585099_5.png" ]
397
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Open the downloads<image>Thought: Type \"download\" in the search bar to locate the Downloads section on your device.\nAction: {\"action\": \"TYPE\", \"value\": \"\", \"position\": null}<image>Thought: Click the \"X\" to remove the search term and access the main menu options for downloads.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.92, 0.06]}<image>", "role": "user" }, { "content": "Thought: Press back to return to the previous screen where downloads might be accessible.\nAction: {\"action\": \"PRESS BACK\", \"value\": null, \"position\": null}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/14556377186823585099_4.png", "training_data/AITW/aitw_images/general/14556377186823585099_5.png", "training_data/AITW/aitw_images/general/14556377186823585099_6.png" ]
398
[ { "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 relative coordinates on the screenshot and should be scaled to a range of 0-1.\nTask: Open the downloads<image>Thought: Click the \"X\" to remove the search term and access the main menu options for downloads.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.92, 0.06]}<image>Thought: Press back to return to the previous screen where downloads might be accessible.\nAction: {\"action\": \"PRESS BACK\", \"value\": null, \"position\": null}<image>", "role": "user" }, { "content": "Thought: Scroll upward to reveal the top navigation where downloads may be located.\nAction: {\"action\": \"CLICK\", \"value\": null, \"position\": [0.84, 0.06]}", "role": "assistant" } ]
[ "training_data/AITW/aitw_images/general/14556377186823585099_5.png", "training_data/AITW/aitw_images/general/14556377186823585099_6.png", "training_data/AITW/aitw_images/general/14556377186823585099_7.png" ]
399