Datasets:

neulab

/

Mind2Web_train_llava

like 2

Follow

NeuLab @ LTI/CMU 65

{ "bytes": "", "path": "./images/5f2c3149-1820-4632-bfdb-3cf4615de2cc_325b7d4b-c635-4187-851d-8219f9a98b4c.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Browse dog food for puppies that is rated at least 4 stars and sort the results by price from low to high.\nPrevious actions:\n[textbox] Search Amazon -> TYPE: dog food\n[button] dog food -> CLICK\n[i] -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.016, 0.51, 0.219, 0.528] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/446e3135-8a53-455f-9471-9f6660f6a94d_a811817b-439d-4624-995e-f2151b37a537.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Book a best business class flight ticket from Colombo to New York with a two-day Qatar stopover for 2 adults and 1 child on June 24.\nPrevious actions:\n[textbox] From Autocomplete selection. Enter your place. -> TYPE: colombo" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.06, 0.242, 0.5, 0.28] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/1f128c19-3093-4c49-bfc7-13589aa8e88a_073be97e-382f-4914-9a6e-ba12b35d6460.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Get a Hasbro Hulk action figure manufactured in 1990 with the lowest price + shipping.\nPrevious actions:\n[tab] Character -> CLICK\n[checkbox] Hulk -> CLICK\n[tab] Year Manufactured -> CLICK\n[checkbox] 1990 -> CLICK\n[button] Apply -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.05, 0.27, 0.064, 0.279] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/f4dcbd09-45f3-47b5-a49d-36658d5b50ca_2e19a5e8-4201-46f0-a062-f812e7f06f8a.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Find tickets for Pizza making class in New York next month.\nPrevious actions:\n[svg] -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.692, 0.086, 0.705, 0.096] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/f1f4e4f7-3fb0-4aeb-8123-4882a1f39cb2_b20324e8-7daa-49b4-b79c-fcc6ef95992d.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Find the new arrival of kids graphic t-shirt for 13 years old boy\nPrevious actions:\n[link] KIDS -> CLICK\n[tab] Tops -> CLICK\n[link] UT: Graphic Tees -> CLICK\n[gridcell] Size -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.063, 0.484, 0.139, 0.504] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/c95b0276-a588-4aeb-906f-a3f6f1c205ae_442b42be-0716-4617-8548-d72fbceb218d.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Show me the list of BGG shirts for sale\nPrevious actions:\nNone" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.32, 0.008, 0.393, 0.03] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/03e45ce0-4375-44aa-b57f-cf439ccbe363_540f58be-846a-4639-988b-214d708f6238.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Find the latest news article and send an email about it.\nPrevious actions:\nNone" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.692, 0.092, 0.963, 0.1] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/16e81dcf-9762-4b83-be3e-60a65a15efec_8ff449e2-99a5-48b0-94be-3804c92710c4.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Book a first class vacation travel for 2 adults from San Francisco to New york leavind on april 9 and returning on april 19\nPrevious actions:\n[a] SFO - San Francisco International, CA -> CLICK\n[textbox] To -> TYPE: new york\n[a] NYC - New York, NY -> CLICK\n[combobox] Number of passengers -> SELECT: 2\n[textbox] Depart (date format mm/dd/yyyy) -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.344, 0.401, 0.358, 0.413] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/837989e0-7476-496a-be02-e4c69f1f989b_27977c97-543d-4538-bfb8-ac7679262132.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Find a cheapest one way flight from Mumbai to Dubai on 24 May.\nPrevious actions:\n[span] Dubai, United Arab Emirates -> CLICK\n[svg] -> CLICK\n[tab] One-way -> CLICK\n[generic] Thursday May 18th -> CLICK\n[div] 24 -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.931, 0.112, 0.975, 0.14] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/0dd0b532-7bac-4b3e-b5a4-2ca6b0897af6_20d5f662-5659-41f2-873e-3f1a4a681fe1.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Browse comedy TV shows streaming on HBO Max that was released in 2010.\nPrevious actions:\n[link] What to Watch on HBO Max -> CLICK\n[button] Filter -> CLICK\n[button] Comedy -> CLICK\n[button] TV Shows -> CLICK\n[span] 2010 -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.361, 0.438, 0.639, 0.463] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/b28e6a37-6e9c-46bb-8012-18ca742a521f_1650528f-5ac6-4a51-91a8-e217fa90b7a0.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Explore Durban in Africa and book a flight from Doha for 2 adults and 1 infant, on May 28, book the first flight in economy comfort.\nPrevious actions:\n[button] Explore (current) -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.254, 0.114, 0.284, 0.129] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/1860601a-faef-4dba-8cbb-807ea8434dca_a7c0ac26-f51d-45fc-969f-73a22770dfc0.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Check for travel requirements when travelling from New York to Tokyo as US citizen.\nPrevious actions:\n[tab] Travel Info -> CLICK\n[link] Travel Planning Center -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.055, 0.102, 0.322, 0.131] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/8ab1c90f-52e0-4c07-9d41-d44c75c25c5e_cb7997fb-4091-45ac-b23e-3c4f1e114867.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Search for an automatic grey sports car with the lowest price, gas fuel and free shipping manufactured between 2018 to 2022, compare the top two results and compare photos.\nPrevious actions:\n[menuitem] 2022 -> CLICK\n[button] Back to all categories -> CLICK\n[button] Exterior Color -> CLICK\n[listitem] Gray (5) Gray (5) -> CLICK\n[button] Back to all categories -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.0, 0.296, 0.253, 0.324] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/08a998f9-82f7-48c8-b6ba-72a58d39b457_85e614b4-41b1-43b9-8519-71f86c26641d.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Find all Uniqlo's located in Chicago, IL.\nPrevious actions:\n[link] Store Locator -> CLICK\n[searchbox] SEARCH BY KEYWORD -> CLICK\n[searchbox] SEARCH BY KEYWORD -> TYPE: Chicago" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.821, 0.178, 0.959, 0.213] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/7ec95f61-b8f4-4431-84d2-55d9f7cdd594_fba5ff31-33b7-42ac-81a2-29fd9779dc8c.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: check flight status for a flight from united airlines leaving Belo Horizonte to Buenos Aires on April 9\nPrevious actions:\n[option] Apr 9, 2023 -> CLICK\n[combobox] Airline. Enter to change collapsed list. Selected ... -> CLICK\n[option] United Airlines (UA) -> CLICK\n[combobox] Enter your departing city, airport name, or airpor... -> TYPE: belo horizonte\n[button] Belo Horizonte, MG, BR (CNF - Tancredo Neves) -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.362, 0.295, 0.637, 0.322] in the image\nAnd my action is Action: TYPE\nValue: buenos aires" } ]

{ "bytes": "", "path": "./images/3d76a0fc-7ba9-4a74-ac52-db0cea359553_acaf6392-4605-4e32-a646-d1b3d7675895.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Find the lowest-priced round trip flight with hotel on May 2 from Kathmandu, Nepal KTM to Shanghai, China PVG and return on May 5. Book a double room and check out with the default flights.\nPrevious actions:\n[checkbox] 5 May 2023 -> CLICK\n[button] Search -> CLICK\n[button] Recommended -> CLICK\n[div] Lowest Price -> CLICK\n[svg] -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.842, 0.316, 0.968, 0.338] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/08a0c5cc-41a9-4eb8-b3ea-91d9e26f8bd3_455d1b00-132a-42d9-bb64-a7c94ae46392.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Get directions for the Koa resort campground in Arizona\nPrevious actions:\n[link] CAMPING AT KOA  -> CLICK\n[link] KOA RESORT CAMPGROUNDS  -> CLICK\n[button] Arizona -> CLICK\n[link] Tucson / Lazydays KOA Resort -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.561, 0.151, 0.659, 0.159] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/8f6374b0-36f1-478e-a282-a61849c8174f_fb4c8a9d-fe56-4fa9-9c3f-9c8044e3e2c6.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Add eleven produce items from Aldi, including 1 pack of blueberry, 6 bananas, 1 grape tomato, 1 roma tomato, 1 cilantro, to my cart and have it delivered tomorrow between 10am and 1pm to the default address.\nPrevious actions:\n[button] View Cart. Items in cart: 5 -> CLICK\n[button] View Cart. Items in cart: 5 -> CLICK\n[div] Go to checkout -> CLICK\n[div] -> CLICK\n[button] Confirm address -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.113, 0.481, 0.619, 0.535] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/ca049641-9721-4593-95c5-a47e22365b5b_37e9f402-9fb7-4e3e-a1fe-8756bffbffbb.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Find the cheapest new electric car with highest ev mile range per charge, see the details of the top result.\nPrevious actions:\n[link] Cars for Sale -> HOVER\n[link] Cars for Sale -> CLICK\n[checkbox] New (9,776) -> CLICK\n[heading] Fuel Type -> CLICK\n[checkbox] Electric (175) -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.113, 0.589, 0.296, 0.635] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/a8474730-82eb-4f12-b947-db64ac08660b_4c7017eb-b5a7-45a4-9644-d3b39dfe5c2e.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Find the most recently posted job opening in customer service in Ireland with a remote working option.\nPrevious actions:\n[div] Discover jobs around the world -> CLICK\n[ppc-content] Ireland -> CLICK\n[button] Flexible Workstyles -> CLICK\n[span] -> CLICK\n[button] Category -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.042, 0.139, 0.052, 0.146] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/a67318a4-5049-4ac5-8c6b-c14fc527483f_6911a96a-5cf1-45ea-a4ac-7b020fa68506.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Browse used BWM cars made in 2010 and compare the four cheapest listings\nPrevious actions:\n[select] 1992 -> SELECT: 2010\n[button] Close dialog -> CLICK\n[select] 2024 -> SELECT: 2010\n[combobox] Sort By -> SELECT: Price: Low to High\n[span] -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.42, 0.614, 0.489, 0.63] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/4c4ab02e-a617-490d-96c6-333d7fa82fe9_9d13499c-38c1-4c07-9165-7af7d7dc7bee.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Check the status of my upcoming trip with booking number 123456789 and last name Smith.\nPrevious actions:\n[tab] My Trips -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.06, 0.285, 0.353, 0.317] in the image\nAnd my action is Action: TYPE\nValue: 123456789" } ]

{ "bytes": "", "path": "./images/cf8da12a-e1fd-48b5-8c53-b88f301ca1e0_cde960df-47f6-41cd-bd34-9082cede2dfa.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Search for video game consoles and filter the results to show only those that have Wi-Fi Capability.\nPrevious actions:\n[link] Electronics -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.013, 0.415, 0.173, 0.429] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/9ebd069a-7703-47b5-9c75-53958637e7c0_ba2bbee3-bfcc-4bab-91ef-2fd4893e6c39.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Watch the halftime show from the most recent Super Bowl\nPrevious actions:\n[use] -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.73, 0.038, 0.787, 0.044] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/c175fe34-143b-4a87-a462-e8e9d69d0b95_3cd2999e-b440-48bd-9a23-54e47a560466.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Check flight status from London to New York City for next day.\nPrevious actions:\n[link] Flight status -> CLICK\n[textbox] From , required. -> TYPE: London\n[a] LON - London, United Kingdom -> CLICK\n[textbox] To , required. -> TYPE: New York" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.348, 0.138, 0.641, 0.154] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/712d9d7f-d7ee-4554-affa-133449231ae4_f7af6222-9fbe-4bed-9d34-344c135ddca0.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Find the latest medical audiobook on complementary medicine for under 20 dollars and add the top book to the cart.\nPrevious actions:\n[link] Shop by category∨ -> HOVER" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.177, 0.182, 0.331, 0.191] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/1337eca1-4ca6-4102-964e-e7f6ecebcd0b_347640d1-0b66-485b-9360-cf59dc1ce10c.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Find things to do in Singapore on April 2 that are family friendly and offer free cancellation\nPrevious actions:\n[tab] Things to do -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.069, 0.143, 0.568, 0.183] in the image\nAnd my action is Action: TYPE\nValue: singapore" } ]

{ "bytes": "", "path": "./images/767faaa6-220e-4e6c-ac1d-39b1501c69cf_869e8d1b-c77b-48a8-9a52-9b34eace9019.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Hire a large car with air conditioning for a small group to pick up from Berlin on May 15, 11:30 am, and drop off at the exact location on May 16, 11 am, book with full protection and payment with a credit card.\nPrevious actions:\n[textbox] I need a car in -> TYPE: berlin\n[button] Berlin Berlin Federal State, Germany (City) -> CLICK\n[div] -> CLICK\n[generic] 15 -> CLICK\n[generic] 16 -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.266, 0.827, 0.372, 0.858] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/277bdab6-e924-45dd-ba26-d659d864d671_a69f9ab4-9419-40c0-a22b-d1bad1fd7c55.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Add 2 different flavor dog treats and one dog food with lowest price to cart from Safeway.\nPrevious actions:\n[textbox] Search -> TYPE: dog treats\n[span] dog treats -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.05, 0.156, 0.956, 0.193] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/06497906-8b76-4367-a1b3-ebc8238bd470_f00c7706-22fe-42e0-bc77-2312bea8c3d4.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Choose a lowest priced hybrid SUV car to pick up from Bangkok downtown on May 24 at 9:30 am and drop off on May 28, at 9 am.\nPrevious actions:\n[checkbox] 28 May 2023 -> CLICK\n[combobox] Pick up time -> SELECT: 9:30 AM\n[combobox] Drop off time -> SELECT: 9:00 AM\n[div] Search -> CLICK\n[div] SUVs -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.084, 0.333, 0.331, 0.343] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/240952bd-853b-4653-a358-49c2784cf568_1b7055a0-0d97-4373-acf5-fd2eb15e484b.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Find a train scheduled between Chicago to Los Angeles for the next day, and if a southwest chief train is available, then print the route details.\nPrevious actions:\n[link] SCHEDULES -> CLICK\n[input] -> TYPE: chicago\n[option] Chicago, IL - Union Station (CHI) -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.374, 0.071, 0.553, 0.082] in the image\nAnd my action is Action: TYPE\nValue: los angeles" } ]

{ "bytes": "", "path": "./images/da679cfe-9180-49f7-b8a2-a2b1bd6d8803_14aad43c-c6e4-4205-beaa-94ff9abb394c.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Show me all the artists with smith in their name\nPrevious actions:\nNone" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.531, 0.059, 0.838, 0.09] in the image\nAnd my action is Action: TYPE\nValue: smith" } ]

{ "bytes": "", "path": "./images/2177b546-5718-478b-8535-bc37975b0d0c_aea31efd-c391-4099-a13e-3a9417cca68f.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Book 2 hot seat VIP tickets through email id adelefan@hotmail.com for Adele's concert in Las Vegas on the last weekend of June.\nPrevious actions:\n[link] TICKETS -> CLICK\n[div] Quantity -> CLICK\n[label] 2 -> CLICK\n[button] $3,535/ea -> CLICK\n[textbox] *Email Address -> TYPE: adelefan@hotmail.com" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.477, 0.448, 0.64, 0.479] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/64779409-c396-4779-a574-b219c3b22282_ee6dfe65-aa46-4181-97d8-3c1944f1ba7c.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Find the longest skiing activity in August 10\nPrevious actions:\n[path] -> CLICK\n[button] Next -> CLICK\n[button] Next -> CLICK\n[button] Next -> CLICK\n[gridcell] Thu Aug 10 2023 -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.656, 0.458, 0.78, 0.48] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/627a99f1-b2e4-4ce4-93f5-89e691f8c879_89295463-6730-47b5-9111-ef25b548fe7a.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Find Kevin Durant's bio\nPrevious actions:\n[button] Open Search -> CLICK\n[textbox] Search -> TYPE: Kevin Durant" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.678, 0.07, 0.931, 0.092] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/298c854d-2987-498b-b43e-6a6452fb11c7_349619be-15c9-4731-a1c9-c020a40df044.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Find unique experiences in London of maximum one hour duration which are rated four stars and above\nPrevious actions:\nNone" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.271, 0.217, 0.463, 0.224] in the image\nAnd my action is Action: TYPE\nValue: London" } ]

{ "bytes": "", "path": "./images/6ca55141-9ba6-4ed7-8c07-77df6c24029d_148ed09c-e612-460c-8704-bafa61872edf.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Find the lowest-priced gray-colored round-shaped mirror in stock in Atlanta, Georgia, with a maximum price of 20 dollars.\nPrevious actions:\n[button] Sort Best match -> CLICK\n[radio] Price: low to high -> CLICK\n[button] Sort -> CLICK\n[button] Shape -> CLICK\n[span] -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.95, 0.146, 0.969, 0.156] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/d4d4a01e-47d0-4e54-8bd5-3d85acebb6dd_b47f5256-ae9f-4c40-8a64-189b47fe6849.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Find the latest action tv shows which are rated fresh by both the audience and the rotten tomatoes\nPrevious actions:\n[link] TV SHOWS -> CLICK\n[span] Sort: Most popular -> CLICK\n[span] Newest -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.204, 0.409, 0.274, 0.438] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/94bd2a0f-fe08-4420-a0e8-d48248630b6e_9df6a79e-671a-40f0-bc71-b7394d96f511.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Find the Leaderboard for the top 10 fantasy Basketball players for the Rotisserie challenge.\nPrevious actions:\n[link] Fantasy . -> CLICK\n[link] Fantasy Basketball -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.169, 0.148, 0.221, 0.165] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/cc643c1f-e73e-4e9f-abe9-5ee2d3b0a0dd_08603353-f42f-498c-8b7e-74aceb336815.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Check the lowest price for a white 2017 certified pre owned Lexus Rx350 near 90012.\nPrevious actions:\n[LabelText] Certified Pre-Owned -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.249, 0.144, 0.358, 0.167] in the image\nAnd my action is Action: SELECT\nValue: Lexus" } ]

{ "bytes": "", "path": "./images/87b23bf1-3c23-45a6-82bf-5a535c7cdc2f_eac5aaaf-f188-42f8-a20d-90b6695ad4b9.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Book the cheapest long-term car rental with a fuel plan, GPS, and satellite radio from June 29 to July 28, pick up at 9 am at Orlando airport and dropping off at the same location.\nPrevious actions:\n[span] Orlando Intl Airport -> CLICK\n[textbox] mm/dd/yyyy -> CLICK\n[link] 29 -> CLICK\n[link] 28 -> CLICK\n[combobox] Pick Up Time -> SELECT: 9:00 AM" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.701, 0.154, 0.902, 0.187] in the image\nAnd my action is Action: SELECT\nValue: 9 00 AM" } ]

{ "bytes": "", "path": "./images/1f128c19-3093-4c49-bfc7-13589aa8e88a_6651b25d-7ba0-4963-a7d0-3211b2eb79b6.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Get a Hasbro Hulk action figure manufactured in 1990 with the lowest price + shipping.\nPrevious actions:\n[checkbox] Hulk -> CLICK\n[tab] Year Manufactured -> CLICK\n[checkbox] 1990 -> CLICK\n[button] Apply -> CLICK\n[input] -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.831, 0.062, 0.906, 0.076] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/b2dd00ff-3f09-46bf-a043-7b996f3815da_5edb6bac-5471-482d-904a-09635fe4ee2d.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Search for holiday campground in Alaska.\nPrevious actions:\n[link] CAMPING AT KOA  -> CLICK\n[link] KOA HOLIDAY CAMPGROUNDS  -> CLICK\n[button] Alaska -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.298, 0.137, 0.481, 0.145] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/6fd2fbb1-9f43-4c46-be42-569a892648e6_5aaf3df7-66f5-437f-8051-2cb596fbdcde.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Add a smartwatch to the cart that is compatible with an iPhone.\nPrevious actions:\nNone" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.28, 0.009, 0.651, 0.04] in the image\nAnd my action is Action: TYPE\nValue: smartwatch" } ]

{ "bytes": "", "path": "./images/2eaf60d5-d2af-4f50-a3ad-5cc9132a770a_da2be31e-ad27-4939-b2f1-d7ad426c736d.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Buy the cheapest used women's running shoes available in a size 5\nPrevious actions:\n[menuitem] Running Shoes -> CLICK\n[label] 5 (10) -> CLICK\n[combobox] Sort By: -> SELECT: Price: Low - High\n[img] -> CLICK\n[radio] 5 -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.5, 0.574, 0.62, 0.604] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/aaade1d2-3a7b-4373-be01-ef98ed86288e_bda0171c-5a26-4bdc-83a3-b7f80d13498e.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Find the cheapest new trash can with automated lid that is under $60 and available for Buy It Now.\nPrevious actions:\nNone" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.214, 0.033, 0.652, 0.058] in the image\nAnd my action is Action: TYPE\nValue: trash can automatic lid" } ]

{ "bytes": "", "path": "./images/27724810-0bc8-446a-a2f4-b53a87e190df_63a067c3-55b3-4dfd-934a-c0bec0d8dccf.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Find the location and operating hours of the nearest CVS pharmacy to zip code 90028\nPrevious actions:\nNone" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.887, 0.013, 0.911, 0.028] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/a3edc9c7-397d-478b-a0b9-0ffa0897acca_71b5600a-25ce-4afd-9929-242be8f64d5f.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Search for person named John Smith in New York.\nPrevious actions:\nNone" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.188, 0.147, 0.26, 0.155] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/a3edc9c7-397d-478b-a0b9-0ffa0897acca_6fc4eaa0-463a-4839-87d9-6df70024abef.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Search for person named John Smith in New York.\nPrevious actions:\n[link] Find People -> CLICK\n[input] -> TYPE: John\n[input] -> TYPE: Smith\n[input] -> TYPE: New York" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.805, 0.282, 0.852, 0.305] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/2879afa9-05f9-4d97-bbfe-f95f5d665174_8e5a056f-17cf-401a-8338-09bd7aad3e3e.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Play the Wakanda Forever trailer\nPrevious actions:\n[link] Navigate to on-demand -> HOVER\n[link] select to navigate to Genres -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.031, 0.332, 0.152, 0.453] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/7219145b-ada1-480e-a8ea-f4fe164ca84a_6c9158f0-6c3b-4ef9-9e89-c09c74149da8.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Book a one way ticket from Faro to Bournemouth leaving on 30th April, with a value fare, for a passenger named Mr. Joe Bloggs. Don't select now and only take 1 small bag onboard.\nPrevious actions:\n[div] -> CLICK\n[button] Search -> CLICK\n[button] Select -> CLICK\n[button] Continue for $ 148 . 16 -> CLICK\n[button] Continue with Value fare -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.123, 0.783, 0.877, 0.839] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/5a15ea92-571d-4a45-ad18-5b4eddbc70a6_56dff79d-4441-4435-a096-71da4f343a09.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Get the lowest priced plus size one piece active swimsuit in color black.\nPrevious actions:\n[link] Shop by Category -> CLICK\n[link] Women -> HOVER" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.312, 0.311, 0.36, 0.32] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/4d311ce8-c660-43dc-bde9-d06e85f6a12c_2895a313-9711-4a8b-9467-88df99b6dd48.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Book a room for a business meeting from Apr 30 to May 1 with 7 people within 5 miles of New Delhi. The hotel must have a business center and offer high-speed internet, select the nearest hotel.\nPrevious actions:\n[link] Business Meetings  -> CLICK\n[textbox] Destination -> TYPE: new delhi" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.041, 0.225, 0.335, 0.261] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/16e81dcf-9762-4b83-be3e-60a65a15efec_5541dc40-3d3c-4624-8e10-e78d62032873.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Book a first class vacation travel for 2 adults from San Francisco to New york leavind on april 9 and returning on april 19\nPrevious actions:\n[button] 􎀺 View Calendar: To navigate the calendar, use the... -> CLICK\n[link] 9 -> CLICK\n[button] 􎀺 View Calendar: To navigate the calendar, use the... -> CLICK\n[link] 19 -> CLICK\n[combobox] Fare preference -> SELECT: First" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.636, 0.322, 0.875, 0.337] in the image\nAnd my action is Action: CLICK" } ]

{ "bytes": "", "path": "./images/58f811fd-0f17-430b-befc-885605e13e41_45d235df-68bb-4808-aab1-97e2fdf6fee8.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: Cancel an order with order number X123456 of Harry Potter Box Set as not available at address.\nPrevious actions:\n[link] Cancel my order -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.235, 0.25, 0.759, 0.278] in the image\nAnd my action is Action: TYPE\nValue: James Smith" } ]

{ "bytes": "", "path": "./images/a531a379-7770-4b72-a8cc-ee19419536cb_b477c115-47bc-4eeb-8d34-b128c643d648.png" }

[ { "from": "human", "value": "<image>\nImagine that you are imitating humans doing web navigation for a task step by step. At each stage, you can see the webpage like humans by a screenshot and know the previous actions before the current step decided by yourself through recorded history. You need to decide on the first following action to take. You can click an element with the mouse, select an option, or type text with the keyboard. (For your understanding, they are like the click(), select_option() and type() functions in playwright respectively) One next step means one operation within the three.\nPlease pinpoint the bounding box coordinates of the element’s region that I should click in order to complete the following instruction: search for art galleries in midtown\nPrevious actions:\n[link] attractions. -> CLICK" }, { "from": "gpt", "value": "For my next step, the target element of my action is bounded by the box [0.145, 0.238, 0.395, 0.261] in the image\nAnd my action is Action: CLICK" } ]