From 6952b45de4f3df7972b806a885d30a412ed7f641 Mon Sep 17 00:00:00 2001 From: Timothyxxx <384084775@qq.com> Date: Fri, 26 Jan 2024 23:30:04 +0800 Subject: [PATCH] Improve on agent and tasks configs --- desktop_env/controllers/python.py | 2 +- .../13584542-872b-42d8-b299-866967b5c3ef.json | 10 ++++ .../23393935-50c7-4a86-aeea-2b78fd089c5c.json | 10 ++++ .../28cc3b7e-b194-4bc9-8353-d04c0f4d56d2.json | 13 ++++- .../3ce045a0-877b-42aa-8d2c-b4a863336ab8.json | 13 ++++- .../43c2d64c-bab5-4dcb-a30c-b888321c319a.json | 15 +++++- .../5ea617a3-0e86-4ba6-aab2-dac9aa2e8d57.json | 10 ++++ .../765d2b74-88a7-4d50-bf51-34e4106fd24a.json | 13 ++++- .../7688b85f-87a4-4e4a-b2f8-f3d6c3f29b82.json | 10 ++++ .../94d95f96-9699-4208-98ba-3c3119edf9c2.json | 13 ++++- .../a462a795-fdc7-4b23-b689-e8b6df786b78.json | 10 ++++ .../a4d98375-215b-4a4d-aee9-3d4370fccc41.json | 12 +++++ .../ae039631-2b12-4637-84f6-c67d51511be3.json | 10 ++++ .../b6781586-6346-41cd-935a-a6b1487918fc.json | 10 ++++ .../bedcedc4-4d72-425e-ad62-21960b11fe0d.json | 13 ++++- .../ddc75b62-7311-4af8-bfb3-859558542b36.json | 12 +++++ .../e0df059f-28a6-4169-924f-b9623e7184cc.json | 10 ++++ .../ec4e3f68-9ea4-4c18-a5c9-69f89d1178b3.json | 10 ++++ .../f9be0997-4b7c-45c5-b05c-4612b44a6118.json | 10 ++++ .../fe41f596-a71b-4c2f-9b2f-9dcd40b568c3.json | 54 +++++++++++-------- .../215dfd39-f493-4bc3-a027-8a97d72c61bf.json | 10 ++++ .../386dbd0e-0241-4a0a-b6a2-6704fba26b1c.json | 10 ++++ .../59f21cfb-0120-4326-b255-a5b827b38967.json | 10 ++++ .../8ba5ae7a-5ae5-4eab-9fcc-5dd4fe3abf89.json | 10 ++++ .../8d9fd4e2-6fdb-46b0-b9b9-02f06495c62f.json | 27 +++++++++- .../8f080098-ddb1-424c-b438-4e96e5e4786e.json | 10 ++++ .../9195653c-f4aa-453d-aa95-787f6ccfaae9.json | 10 ++++ .../a5bbbcd5-b398-4c91-83d4-55e1e31bbb81.json | 10 ++++ .../aa4b5023-aef6-4ed9-bdc9-705f59ab9ad6.json | 10 ++++ .../bba3381f-b5eb-4439-bd9e-80c22218d5a7.json | 10 ++++ .../d06f0d4d-2cd5-4ede-8de9-598629438c6e.json | 10 ++++ .../efcf0d81-0835-4880-b2fd-d866e8bc2294.json | 10 ++++ .../f3977615-2b45-4ac5-8bba-80c17dbe2a37.json | 10 ++++ .../fba2c100-79e8-42df-ae74-b592418d54f4.json | 22 ++++++++ mm_agents/gpt_4v_agent.py | 12 ++--- mm_agents/prompts.py | 30 +++++++---- 36 files changed, 425 insertions(+), 46 deletions(-) diff --git a/desktop_env/controllers/python.py b/desktop_env/controllers/python.py index a4b2abb..1f0d32e 100644 --- a/desktop_env/controllers/python.py +++ b/desktop_env/controllers/python.py @@ -11,7 +11,7 @@ logger = logging.getLogger("desktopenv.pycontroller") class PythonController: - def __init__(self, vm_ip: str, pkgs_prefix: str = "import pyautogui; {command}"): + def __init__(self, vm_ip: str, pkgs_prefix: str = "import pyautogui; import time; {command}"): self.vm_ip = vm_ip self.http_server = f"http://{vm_ip}:5000" self.pkgs_prefix = pkgs_prefix # fixme: this is a hacky way to execute python commands. fix it and combine it with installation of packages diff --git a/evaluation_examples/examples/os/13584542-872b-42d8-b299-866967b5c3ef.json b/evaluation_examples/examples/os/13584542-872b-42d8-b299-866967b5c3ef.json index 4e08188..da14f2c 100644 --- a/evaluation_examples/examples/os/13584542-872b-42d8-b299-866967b5c3ef.json +++ b/evaluation_examples/examples/os/13584542-872b-42d8-b299-866967b5c3ef.json @@ -5,6 +5,16 @@ "source": "https://superuser.com/questions/72176/linux-set-default-terminal-size-and-screen-position", "trajectory": "trajectories/", "config": [ + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } + } ], "related_apps": [ "os" diff --git a/evaluation_examples/examples/os/23393935-50c7-4a86-aeea-2b78fd089c5c.json b/evaluation_examples/examples/os/23393935-50c7-4a86-aeea-2b78fd089c5c.json index 654a749..239fd23 100644 --- a/evaluation_examples/examples/os/23393935-50c7-4a86-aeea-2b78fd089c5c.json +++ b/evaluation_examples/examples/os/23393935-50c7-4a86-aeea-2b78fd089c5c.json @@ -80,6 +80,16 @@ } ] } + }, + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } } ], "related_apps": [ diff --git a/evaluation_examples/examples/os/28cc3b7e-b194-4bc9-8353-d04c0f4d56d2.json b/evaluation_examples/examples/os/28cc3b7e-b194-4bc9-8353-d04c0f4d56d2.json index a88aaf3..a5bc881 100644 --- a/evaluation_examples/examples/os/28cc3b7e-b194-4bc9-8353-d04c0f4d56d2.json +++ b/evaluation_examples/examples/os/28cc3b7e-b194-4bc9-8353-d04c0f4d56d2.json @@ -4,7 +4,18 @@ "instruction": "The volume of my system is too small. Can you help me turn up to the max volume?", "source": "https://help.ubuntu.com/lts/ubuntu-help/sound-volume.html.en", "trajectory": "trajectories/", - "config": [], + "config": [ + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } + } + ], "related_apps": [ "os" ], diff --git a/evaluation_examples/examples/os/3ce045a0-877b-42aa-8d2c-b4a863336ab8.json b/evaluation_examples/examples/os/3ce045a0-877b-42aa-8d2c-b4a863336ab8.json index 99df0f7..6cb0bea 100644 --- a/evaluation_examples/examples/os/3ce045a0-877b-42aa-8d2c-b4a863336ab8.json +++ b/evaluation_examples/examples/os/3ce045a0-877b-42aa-8d2c-b4a863336ab8.json @@ -4,7 +4,18 @@ "instruction": "My glasses are broken, and I'm having trouble seeing small things clearly. Could you help me enlarge the text on my screen so it's easier to read?", "source": "https://help.ubuntu.com/lts/ubuntu-help/a11y-font-size.html.en", "trajectory": "trajectories/", - "config": [], + "config": [ + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } + } + ], "related_apps": [ "os" ], diff --git a/evaluation_examples/examples/os/43c2d64c-bab5-4dcb-a30c-b888321c319a.json b/evaluation_examples/examples/os/43c2d64c-bab5-4dcb-a30c-b888321c319a.json index e537a60..bf6a665 100644 --- a/evaluation_examples/examples/os/43c2d64c-bab5-4dcb-a30c-b888321c319a.json +++ b/evaluation_examples/examples/os/43c2d64c-bab5-4dcb-a30c-b888321c319a.json @@ -3,7 +3,18 @@ "snapshot": "os", "instruction": "Could you please help me create a dir named 'test' in the root directory of current computer?", "source": "https://ubuntu.com/tutorials/command-line-for-beginners#4-creating-folders-and-files", - "config": [], + "config": [ + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } + } + ], "trajectory": "trajectories/", "related_apps": [ "os" @@ -12,7 +23,7 @@ "func": "exact_match", "result": { "type": "vm_command_line", - "command": "[ -d '/home/user/Desktop/test' ] && echo 'Directory exists.' || echo 'Directory does not exist.'", + "command": "[ -d '/test' ] && echo 'Directory exists.' || echo 'Directory does not exist.'", "shell": true }, "expected": { diff --git a/evaluation_examples/examples/os/5ea617a3-0e86-4ba6-aab2-dac9aa2e8d57.json b/evaluation_examples/examples/os/5ea617a3-0e86-4ba6-aab2-dac9aa2e8d57.json index 7ce6f21..d51c886 100644 --- a/evaluation_examples/examples/os/5ea617a3-0e86-4ba6-aab2-dac9aa2e8d57.json +++ b/evaluation_examples/examples/os/5ea617a3-0e86-4ba6-aab2-dac9aa2e8d57.json @@ -22,6 +22,16 @@ "command": "mv ~/poster_party_night.webp ~/.local/share/Trash/files/", "shell": true } + }, + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } } ], "related_apps": [ diff --git a/evaluation_examples/examples/os/765d2b74-88a7-4d50-bf51-34e4106fd24a.json b/evaluation_examples/examples/os/765d2b74-88a7-4d50-bf51-34e4106fd24a.json index 690cdc8..98f51bd 100644 --- a/evaluation_examples/examples/os/765d2b74-88a7-4d50-bf51-34e4106fd24a.json +++ b/evaluation_examples/examples/os/765d2b74-88a7-4d50-bf51-34e4106fd24a.json @@ -3,7 +3,18 @@ "snapshot": "os", "instruction": "Can you help me delete the \"test\" file on my desktop?", "source": "https://help.ubuntu.com/lts/ubuntu-help/files-delete.html.en", - "config": [], + "config": [ + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } + } + ], "trajectory": "trajectories/", "related_apps": [ "os" diff --git a/evaluation_examples/examples/os/7688b85f-87a4-4e4a-b2f8-f3d6c3f29b82.json b/evaluation_examples/examples/os/7688b85f-87a4-4e4a-b2f8-f3d6c3f29b82.json index a3b6035..be8d79e 100644 --- a/evaluation_examples/examples/os/7688b85f-87a4-4e4a-b2f8-f3d6c3f29b82.json +++ b/evaluation_examples/examples/os/7688b85f-87a4-4e4a-b2f8-f3d6c3f29b82.json @@ -17,7 +17,17 @@ "command": "echo 'password' | sudo -S chmod 777 ~/Desktop/todo.txt && sudo chmod 777 ~/Desktop/done", "shell": true } + }, + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] } + } ], "trajectory": "trajectories/", "related_apps": [ diff --git a/evaluation_examples/examples/os/94d95f96-9699-4208-98ba-3c3119edf9c2.json b/evaluation_examples/examples/os/94d95f96-9699-4208-98ba-3c3119edf9c2.json index 90249d0..027b7e2 100644 --- a/evaluation_examples/examples/os/94d95f96-9699-4208-98ba-3c3119edf9c2.json +++ b/evaluation_examples/examples/os/94d95f96-9699-4208-98ba-3c3119edf9c2.json @@ -4,7 +4,18 @@ "instruction": "I want to install Spotify on my current system. Could you please help me?", "source": "https://help.ubuntu.com/lts/ubuntu-help/addremove-install.html.en", "trajectory": "trajectories/", - "config": [], + "config": [ + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } + } + ], "related_apps": [ "os" ], diff --git a/evaluation_examples/examples/os/a462a795-fdc7-4b23-b689-e8b6df786b78.json b/evaluation_examples/examples/os/a462a795-fdc7-4b23-b689-e8b6df786b78.json index 72d0454..f043475 100644 --- a/evaluation_examples/examples/os/a462a795-fdc7-4b23-b689-e8b6df786b78.json +++ b/evaluation_examples/examples/os/a462a795-fdc7-4b23-b689-e8b6df786b78.json @@ -10,6 +10,16 @@ "command": "echo password | sudo -S su - charles", "shell": true } + }, + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } } ], "trajectory": "trajectories/", diff --git a/evaluation_examples/examples/os/a4d98375-215b-4a4d-aee9-3d4370fccc41.json b/evaluation_examples/examples/os/a4d98375-215b-4a4d-aee9-3d4370fccc41.json index 3675d05..750fea6 100644 --- a/evaluation_examples/examples/os/a4d98375-215b-4a4d-aee9-3d4370fccc41.json +++ b/evaluation_examples/examples/os/a4d98375-215b-4a4d-aee9-3d4370fccc41.json @@ -4,6 +4,18 @@ "instruction": "I want to have my computer automatically locked after I leaved. Can you help me?", "source": "https://help.ubuntu.com/lts/ubuntu-help/privacy-screen-lock.html.en", "trajectory": "trajectories/", + "config": [ + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } + } + ], "related_apps": [ "os" ], diff --git a/evaluation_examples/examples/os/ae039631-2b12-4637-84f6-c67d51511be3.json b/evaluation_examples/examples/os/ae039631-2b12-4637-84f6-c67d51511be3.json index c82d0b0..9c97d21 100644 --- a/evaluation_examples/examples/os/ae039631-2b12-4637-84f6-c67d51511be3.json +++ b/evaluation_examples/examples/os/ae039631-2b12-4637-84f6-c67d51511be3.json @@ -9,6 +9,16 @@ "parameters": { "command": "xdg-settings set default-web-browser firefox.desktop" } + }, + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } } ], "trajectory": "trajectories/", diff --git a/evaluation_examples/examples/os/b6781586-6346-41cd-935a-a6b1487918fc.json b/evaluation_examples/examples/os/b6781586-6346-41cd-935a-a6b1487918fc.json index 46964d3..329673c 100644 --- a/evaluation_examples/examples/os/b6781586-6346-41cd-935a-a6b1487918fc.json +++ b/evaluation_examples/examples/os/b6781586-6346-41cd-935a-a6b1487918fc.json @@ -5,6 +5,16 @@ "source": "https://help.ubuntu.com/lts/ubuntu-help/clock-timezone.html.en", "trajectory": "trajectories/", "config": [ + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } + } ], "related_apps": [ "os" diff --git a/evaluation_examples/examples/os/bedcedc4-4d72-425e-ad62-21960b11fe0d.json b/evaluation_examples/examples/os/bedcedc4-4d72-425e-ad62-21960b11fe0d.json index 3061ba6..6c67c0c 100644 --- a/evaluation_examples/examples/os/bedcedc4-4d72-425e-ad62-21960b11fe0d.json +++ b/evaluation_examples/examples/os/bedcedc4-4d72-425e-ad62-21960b11fe0d.json @@ -4,7 +4,18 @@ "instruction": "Could you set the 'Dim screen when inactive' to on in setting?", "source": "https://www.youtube.com/watch?v=D4WyNjt_hbQ&t=2s", "trajectory": "trajectories/", - "config": [], + "config": [ + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } + } + ], "related_apps": [ "os" ], diff --git a/evaluation_examples/examples/os/ddc75b62-7311-4af8-bfb3-859558542b36.json b/evaluation_examples/examples/os/ddc75b62-7311-4af8-bfb3-859558542b36.json index c54f368..662de63 100644 --- a/evaluation_examples/examples/os/ddc75b62-7311-4af8-bfb3-859558542b36.json +++ b/evaluation_examples/examples/os/ddc75b62-7311-4af8-bfb3-859558542b36.json @@ -4,6 +4,18 @@ "instruction": "I want to uninstall the Mahjongg on my system. Can you help me?", "source": "https://help.ubuntu.com/lts/ubuntu-help/addremove-remove.html.en", "trajectory": "trajectories/", + "config": [ + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } + } + ], "related_apps": [ "os" ], diff --git a/evaluation_examples/examples/os/e0df059f-28a6-4169-924f-b9623e7184cc.json b/evaluation_examples/examples/os/e0df059f-28a6-4169-924f-b9623e7184cc.json index c9e41d8..3c54c5a 100644 --- a/evaluation_examples/examples/os/e0df059f-28a6-4169-924f-b9623e7184cc.json +++ b/evaluation_examples/examples/os/e0df059f-28a6-4169-924f-b9623e7184cc.json @@ -11,6 +11,16 @@ "command": "echo 'password' | sudo -S mkdir ~/Desktop/todo_list_Jan_1", "shell": true } + }, + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } } ], "related_apps": [ diff --git a/evaluation_examples/examples/os/ec4e3f68-9ea4-4c18-a5c9-69f89d1178b3.json b/evaluation_examples/examples/os/ec4e3f68-9ea4-4c18-a5c9-69f89d1178b3.json index 1b4a330..30daecb 100644 --- a/evaluation_examples/examples/os/ec4e3f68-9ea4-4c18-a5c9-69f89d1178b3.json +++ b/evaluation_examples/examples/os/ec4e3f68-9ea4-4c18-a5c9-69f89d1178b3.json @@ -21,6 +21,16 @@ "parameters": { "command": "gsettings set org.gnome.shell favorite-apps \"['thunderbird.desktop', 'vim.desktop', 'google-chrome.desktop']\"" } + }, + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } } ], "trajectory": "trajectories/", diff --git a/evaluation_examples/examples/os/f9be0997-4b7c-45c5-b05c-4612b44a6118.json b/evaluation_examples/examples/os/f9be0997-4b7c-45c5-b05c-4612b44a6118.json index 96632aa..7a07888 100644 --- a/evaluation_examples/examples/os/f9be0997-4b7c-45c5-b05c-4612b44a6118.json +++ b/evaluation_examples/examples/os/f9be0997-4b7c-45c5-b05c-4612b44a6118.json @@ -10,7 +10,17 @@ "command": "gsettings set org.gnome.desktop.notifications show-banners true", "shell": true } + }, + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] } + } ], "trajectory": "trajectories/", "related_apps": [ diff --git a/evaluation_examples/examples/os/fe41f596-a71b-4c2f-9b2f-9dcd40b568c3.json b/evaluation_examples/examples/os/fe41f596-a71b-4c2f-9b2f-9dcd40b568c3.json index c613775..79d8618 100644 --- a/evaluation_examples/examples/os/fe41f596-a71b-4c2f-9b2f-9dcd40b568c3.json +++ b/evaluation_examples/examples/os/fe41f596-a71b-4c2f-9b2f-9dcd40b568c3.json @@ -1,24 +1,36 @@ { - "id": "fe41f596-a71b-4c2f-9b2f-9dcd40b568c3", - "snapshot": "os", - "instruction": "I want to see the battery percentage. Can you help me display it on my screen?", - "source": "https://help.ubuntu.com/lts/ubuntu-help/power-percentage.html.en", - "trajectory": "trajectories/", - "related_apps": [ - "os" - ], - "evaluator": { - "func": "exact_match", - "result": { - "type": "vm_command_line", - "command": "gsettings get org.gnome.desktop.interface show-battery-percentage", - "shell": true - }, - "expected": { - "type": "rule", - "rules":{ - "expected": "true\n" - } + "id": "fe41f596-a71b-4c2f-9b2f-9dcd40b568c3", + "snapshot": "os", + "instruction": "I want to see the battery percentage. Can you help me display it on my screen?", + "source": "https://help.ubuntu.com/lts/ubuntu-help/power-percentage.html.en", + "trajectory": "trajectories/", + "config": [ + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] } } - } \ No newline at end of file + ], + "related_apps": [ + "os" + ], + "evaluator": { + "func": "exact_match", + "result": { + "type": "vm_command_line", + "command": "gsettings get org.gnome.desktop.interface show-battery-percentage", + "shell": true + }, + "expected": { + "type": "rule", + "rules": { + "expected": "true\n" + } + } + } +} \ No newline at end of file diff --git a/evaluation_examples/examples/vlc/215dfd39-f493-4bc3-a027-8a97d72c61bf.json b/evaluation_examples/examples/vlc/215dfd39-f493-4bc3-a027-8a97d72c61bf.json index 21f4222..afaa063 100644 --- a/evaluation_examples/examples/vlc/215dfd39-f493-4bc3-a027-8a97d72c61bf.json +++ b/evaluation_examples/examples/vlc/215dfd39-f493-4bc3-a027-8a97d72c61bf.json @@ -9,6 +9,16 @@ "parameters": { "command": "vlc" } + }, + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } } ], "trajectory": "trajectories/", diff --git a/evaluation_examples/examples/vlc/386dbd0e-0241-4a0a-b6a2-6704fba26b1c.json b/evaluation_examples/examples/vlc/386dbd0e-0241-4a0a-b6a2-6704fba26b1c.json index 75dff19..d2d5dec 100644 --- a/evaluation_examples/examples/vlc/386dbd0e-0241-4a0a-b6a2-6704fba26b1c.json +++ b/evaluation_examples/examples/vlc/386dbd0e-0241-4a0a-b6a2-6704fba26b1c.json @@ -64,6 +64,16 @@ "import pyautogui; pyautogui.press('f11');" ] } + }, + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } } ], "trajectory": "trajectories/", diff --git a/evaluation_examples/examples/vlc/59f21cfb-0120-4326-b255-a5b827b38967.json b/evaluation_examples/examples/vlc/59f21cfb-0120-4326-b255-a5b827b38967.json index 14ce3cd..fa35f68 100644 --- a/evaluation_examples/examples/vlc/59f21cfb-0120-4326-b255-a5b827b38967.json +++ b/evaluation_examples/examples/vlc/59f21cfb-0120-4326-b255-a5b827b38967.json @@ -20,6 +20,16 @@ "parameters": { "command": "vlc" } + }, + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } } ], "trajectory": "trajectories/", diff --git a/evaluation_examples/examples/vlc/8ba5ae7a-5ae5-4eab-9fcc-5dd4fe3abf89.json b/evaluation_examples/examples/vlc/8ba5ae7a-5ae5-4eab-9fcc-5dd4fe3abf89.json index d59d628..66b23fd 100644 --- a/evaluation_examples/examples/vlc/8ba5ae7a-5ae5-4eab-9fcc-5dd4fe3abf89.json +++ b/evaluation_examples/examples/vlc/8ba5ae7a-5ae5-4eab-9fcc-5dd4fe3abf89.json @@ -9,6 +9,16 @@ "parameters": { "command": "vlc" } + }, + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } } ], "trajectory": "trajectories/", diff --git a/evaluation_examples/examples/vlc/8d9fd4e2-6fdb-46b0-b9b9-02f06495c62f.json b/evaluation_examples/examples/vlc/8d9fd4e2-6fdb-46b0-b9b9-02f06495c62f.json index 8c2fdb2..636eb25 100644 --- a/evaluation_examples/examples/vlc/8d9fd4e2-6fdb-46b0-b9b9-02f06495c62f.json +++ b/evaluation_examples/examples/vlc/8d9fd4e2-6fdb-46b0-b9b9-02f06495c62f.json @@ -4,10 +4,35 @@ "instruction": "Can you make the video fill up the whole screen? It's a bit too small to see right now, and I'd like to see it better.", "source": "https://www.youtube.com/watch?v=XHprwDJ0-fU&t=436s", "config": [ + { + "type": "download", + "parameters": { + "files": [ + { + "url": "https://drive.usercontent.google.com/download?id=13OhDRuINzQ-w0qmzdGJ8ZSWnKyKnTRje&export=download&authuser=0&confirm=t&uuid=39cf0297-23eb-47df-a642-fd6c2f981eff&at=APZUnTUR83eI1tenMxVdyFxNeBo7:1706275408621", + "path": "Desktop/Gen 2.mp4" + } + ] + } + }, { "type": "launch", "parameters": { - "command": "vlc" + "command": [ + "vlc", + "--start-time=15", + "Desktop/Gen 2.mp4" + ] + } + }, + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] } } ], diff --git a/evaluation_examples/examples/vlc/8f080098-ddb1-424c-b438-4e96e5e4786e.json b/evaluation_examples/examples/vlc/8f080098-ddb1-424c-b438-4e96e5e4786e.json index a39bd31..7b521e8 100644 --- a/evaluation_examples/examples/vlc/8f080098-ddb1-424c-b438-4e96e5e4786e.json +++ b/evaluation_examples/examples/vlc/8f080098-ddb1-424c-b438-4e96e5e4786e.json @@ -24,6 +24,16 @@ "Desktop/Baby Justin Bieber.mp4" ] } + }, + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } } ], "trajectory": "trajectories/", diff --git a/evaluation_examples/examples/vlc/9195653c-f4aa-453d-aa95-787f6ccfaae9.json b/evaluation_examples/examples/vlc/9195653c-f4aa-453d-aa95-787f6ccfaae9.json index 73d0fd2..3345bf0 100644 --- a/evaluation_examples/examples/vlc/9195653c-f4aa-453d-aa95-787f6ccfaae9.json +++ b/evaluation_examples/examples/vlc/9195653c-f4aa-453d-aa95-787f6ccfaae9.json @@ -9,6 +9,16 @@ "parameters": { "command": "vlc" } + }, + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } } ], "trajectory": "trajectories/", diff --git a/evaluation_examples/examples/vlc/a5bbbcd5-b398-4c91-83d4-55e1e31bbb81.json b/evaluation_examples/examples/vlc/a5bbbcd5-b398-4c91-83d4-55e1e31bbb81.json index ff39f72..5b67718 100644 --- a/evaluation_examples/examples/vlc/a5bbbcd5-b398-4c91-83d4-55e1e31bbb81.json +++ b/evaluation_examples/examples/vlc/a5bbbcd5-b398-4c91-83d4-55e1e31bbb81.json @@ -9,6 +9,16 @@ "parameters": { "command": "vlc" } + }, + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } } ], "trajectory": "trajectories/", diff --git a/evaluation_examples/examples/vlc/aa4b5023-aef6-4ed9-bdc9-705f59ab9ad6.json b/evaluation_examples/examples/vlc/aa4b5023-aef6-4ed9-bdc9-705f59ab9ad6.json index f525f96..e472f97 100644 --- a/evaluation_examples/examples/vlc/aa4b5023-aef6-4ed9-bdc9-705f59ab9ad6.json +++ b/evaluation_examples/examples/vlc/aa4b5023-aef6-4ed9-bdc9-705f59ab9ad6.json @@ -20,6 +20,16 @@ "parameters": { "command": ["vlc", "Desktop/flipped_1984_Apple_Macintosh_Commercial.mp4"] } + }, + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } } ], "trajectory": "trajectories/", diff --git a/evaluation_examples/examples/vlc/bba3381f-b5eb-4439-bd9e-80c22218d5a7.json b/evaluation_examples/examples/vlc/bba3381f-b5eb-4439-bd9e-80c22218d5a7.json index a5cd48d..b6c397c 100644 --- a/evaluation_examples/examples/vlc/bba3381f-b5eb-4439-bd9e-80c22218d5a7.json +++ b/evaluation_examples/examples/vlc/bba3381f-b5eb-4439-bd9e-80c22218d5a7.json @@ -9,6 +9,16 @@ "parameters": { "command": "vlc" } + }, + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } } ], "trajectory": "trajectories/", diff --git a/evaluation_examples/examples/vlc/d06f0d4d-2cd5-4ede-8de9-598629438c6e.json b/evaluation_examples/examples/vlc/d06f0d4d-2cd5-4ede-8de9-598629438c6e.json index c29a8bc..a6bcc8f 100644 --- a/evaluation_examples/examples/vlc/d06f0d4d-2cd5-4ede-8de9-598629438c6e.json +++ b/evaluation_examples/examples/vlc/d06f0d4d-2cd5-4ede-8de9-598629438c6e.json @@ -9,6 +9,16 @@ "parameters": { "command": "vlc" } + }, + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } } ], "trajectory": "trajectories/", diff --git a/evaluation_examples/examples/vlc/efcf0d81-0835-4880-b2fd-d866e8bc2294.json b/evaluation_examples/examples/vlc/efcf0d81-0835-4880-b2fd-d866e8bc2294.json index 3b07f3c..b07c6de 100644 --- a/evaluation_examples/examples/vlc/efcf0d81-0835-4880-b2fd-d866e8bc2294.json +++ b/evaluation_examples/examples/vlc/efcf0d81-0835-4880-b2fd-d866e8bc2294.json @@ -26,6 +26,16 @@ "Desktop/Interstellar Movie - Official Trailer.mp4" ] } + }, + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } } ], "trajectory": "trajectories/", diff --git a/evaluation_examples/examples/vlc/f3977615-2b45-4ac5-8bba-80c17dbe2a37.json b/evaluation_examples/examples/vlc/f3977615-2b45-4ac5-8bba-80c17dbe2a37.json index e1ee77e..3be1e5d 100644 --- a/evaluation_examples/examples/vlc/f3977615-2b45-4ac5-8bba-80c17dbe2a37.json +++ b/evaluation_examples/examples/vlc/f3977615-2b45-4ac5-8bba-80c17dbe2a37.json @@ -9,6 +9,16 @@ "parameters": { "command": "vlc" } + }, + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(960, 540); time.sleep(0.5);" + ] + } } ], "trajectory": "trajectories/", diff --git a/evaluation_examples/examples/vlc/fba2c100-79e8-42df-ae74-b592418d54f4.json b/evaluation_examples/examples/vlc/fba2c100-79e8-42df-ae74-b592418d54f4.json index 0cd51d9..f71069e 100644 --- a/evaluation_examples/examples/vlc/fba2c100-79e8-42df-ae74-b592418d54f4.json +++ b/evaluation_examples/examples/vlc/fba2c100-79e8-42df-ae74-b592418d54f4.json @@ -26,6 +26,28 @@ "Desktop/Interstellar Movie - Official Trailer.mp4" ] } + }, + { + "type": "sleep", + "parameters": { + "seconds": 1 + } + }, + { + "type": "activate_window", + "parameters": { + "window_name": "Interstellar Movie - Interstellar Movie - Official Trailer - VLC media player" + } + }, + { + "type": "execute", + "parameters": { + "command": [ + "python", + "-c", + "import pyautogui; import time; pyautogui.click(500, 500); time.sleep(0.5);" + ] + } } ], "trajectory": "trajectories/", diff --git a/mm_agents/gpt_4v_agent.py b/mm_agents/gpt_4v_agent.py index cc9e9d5..423e554 100644 --- a/mm_agents/gpt_4v_agent.py +++ b/mm_agents/gpt_4v_agent.py @@ -225,8 +225,8 @@ class GPT4v_Agent: }) # Append trajectory - assert len(self.observations) == len(self.actions) and len(self.actions) == len(self.thoughts)\ - , "The number of observations and actions should be the same." + assert len(self.observations) == len(self.actions) and len(self.actions) == len(self.thoughts) \ + , "The number of observations and actions should be the same." if len(self.observations) > self.max_trajectory_length: _observations = self.observations[-self.max_trajectory_length:] @@ -255,7 +255,7 @@ class GPT4v_Agent: { "type": "image_url", "image_url": { - "url": f"data:image/jpeg;base64,{_screenshot}", + "url": f"data:image/png;base64,{_screenshot}", "detail": "high" } } @@ -315,14 +315,14 @@ class GPT4v_Agent: ] }) else: - raise ValueError("Invalid experiment type: " + self.exp) # 1}}} + raise ValueError("Invalid experiment type: " + self.exp) # 1}}} messages.append({ "role": "assistant", "content": [ { "type": "text", - "text": previous_thought.stip() if len(previous_thought)>0 else "No valid action" + "text": previous_thought.strip() if len(previous_thought) > 0 else "No valid action" }, ] }) @@ -436,7 +436,7 @@ class GPT4v_Agent: ] }) else: - raise ValueError("Invalid experiment type: " + self.exp) # 1}}} + raise ValueError("Invalid experiment type: " + self.exp) # 1}}} with open("messages.json", "w") as f: f.write(json.dumps(messages, indent=4)) diff --git a/mm_agents/prompts.py b/mm_agents/prompts.py index 90ce22f..85295de 100644 --- a/mm_agents/prompts.py +++ b/mm_agents/prompts.py @@ -3,8 +3,9 @@ You are an agent which follow my instruction and perform desktop computer tasks You have good knowledge of computer and good internet connection and assume your code will run on a computer for controlling the mouse and keyboard. For each step, you will get an observation of an image, which is the screenshot of the computer screen and you will predict the action of the computer based on the image. -You are required to use `pyautogui` to perform the action, but don't use the `pyautogui.locateCenterOnScreen` function to locate the element you want to operate with since we have no image of the element you want to operate with. -Return one line or multiple lines of python code to perform the action each time, be time efficient. +You are required to use `pyautogui` to perform the action grounded to the observation, but DONOT use the `pyautogui.locateCenterOnScreen` function to locate the element you want to operate with since we have no image of the element you want to operate with. DONOT USE `pyautogui.screenshot()` to make screenshot. +Return one line or multiple lines of python code to perform the action each time, be time efficient. When predicting multiple lines of code, make some small sleep like `time.sleep(0.5);` interval so that the machine could take +You need to to specify the coordinates of by yourself based on your observation of current observation, but you should be careful to ensure that the coordinates are correct. You ONLY need to return the code inside a code block, like this: ```python # your code here @@ -14,6 +15,7 @@ When you think you have to wait for some time, return ```WAIT```; When you think the task can not be done, return ```FAIL```, don't easily say ```FAIL```, try your best to do the task; When you think the task is done, return ```DONE```. +My computer's password is 'password', feel free to use it when you need sudo rights. First give the current screenshot and previous things we did a short reflection, then RETURN ME THE CODE OR SPECIAL CODE I ASKED FOR. NEVER EVER RETURN ME ANYTHING ELSE. """.strip() @@ -267,8 +269,9 @@ You are an agent which follow my instruction and perform desktop computer tasks You have good knowledge of computer and good internet connection and assume your code will run on a computer for controlling the mouse and keyboard. For each step, you will get an observation of the desktop by accessibility tree, which is based on AT-SPI library. And you will predict the action of the computer based on the accessibility tree. -You are required to use `pyautogui` to perform the action, but don't use the `pyautogui.locateCenterOnScreen` function to locate the element you want to operate with since we have no image of the element you want to operate with. -Return one line or multiple lines of python code to perform the action each time, be time efficient. +You are required to use `pyautogui` to perform the action grounded to the observation, but DONOT use the `pyautogui.locateCenterOnScreen` function to locate the element you want to operate with since we have no image of the element you want to operate with. DONOT USE `pyautogui.screenshot()` to make screenshot. +Return one line or multiple lines of python code to perform the action each time, be time efficient. When predicting multiple lines of code, make some small sleep like `time.sleep(0.5);` interval so that the machine could take +You need to to specify the coordinates of by yourself based on your observation of current observation, but you should be careful to ensure that the coordinates are correct. You ONLY need to return the code inside a code block, like this: ```python # your code here @@ -278,6 +281,7 @@ When you think you have to wait for some time, return ```WAIT```; When you think the task can not be done, return ```FAIL```, don't easily say ```FAIL```, try your best to do the task; When you think the task is done, return ```DONE```. +My computer's password is 'password', feel free to use it when you need sudo rights. First give the current screenshot and previous things we did a short reflection, then RETURN ME THE CODE OR SPECIAL CODE I ASKED FOR. NEVER EVER RETURN ME ANYTHING ELSE. """.strip() @@ -532,8 +536,9 @@ You have good knowledge of computer and good internet connection and assume your For each step, you will get an observation of the desktop by 1) a screenshot; and 2) accessibility tree, which is based on AT-SPI library. And you will predict the action of the computer based on the screenshot and accessibility tree. -You are required to use `pyautogui` to perform the action, but don't use the `pyautogui.locateCenterOnScreen` function to locate the element you want to operate with since we have no image of the element you want to operate with. -Return one line or multiple lines of python code to perform the action each time, be time efficient. +You are required to use `pyautogui` to perform the action grounded to the observation, but DONOT use the `pyautogui.locateCenterOnScreen` function to locate the element you want to operate with since we have no image of the element you want to operate with. DONOT USE `pyautogui.screenshot()` to make screenshot. +Return one line or multiple lines of python code to perform the action each time, be time efficient. When predicting multiple lines of code, make some small sleep like `time.sleep(0.5);` interval so that the machine could take +You need to to specify the coordinates of by yourself based on your observation of current observation, but you should be careful to ensure that the coordinates are correct. You ONLY need to return the code inside a code block, like this: ```python # your code here @@ -543,6 +548,7 @@ When you think you have to wait for some time, return ```WAIT```; When you think the task can not be done, return ```FAIL```, don't easily say ```FAIL```, try your best to do the task; When you think the task is done, return ```DONE```. +My computer's password is 'password', feel free to use it when you need sudo rights. First give the current screenshot and previous things we did a short reflection, then RETURN ME THE CODE OR SPECIAL CODE I ASKED FOR. NEVER EVER RETURN ME ANYTHING ELSE. """.strip() @@ -797,7 +803,7 @@ You are an agent which follow my instruction and perform desktop computer tasks You have good knowledge of computer and good internet connection and assume your code will run on a computer for controlling the mouse and keyboard. For each step, you will get an observation of the desktop by 1) a screenshot; and 2) accessibility tree, which is based on AT-SPI library. -You are required to use `pyautogui` to perform the action, but don't use the `pyautogui.locateCenterOnScreen` function to locate the element you want to operate with since we have no image of the element you want to operate with. +You are required to use `pyautogui` to perform the action grounded to the observation, but DONOT use the `pyautogui.locateCenterOnScreen` function to locate the element you want to operate with since we have no image of the element you want to operate with. DONOT USE `pyautogui.screenshot()` to make screenshot. You can replace x, y in the code with the tag of the element you want to operate with. such as: ```python pyautogui.moveTo(tag#3) @@ -806,7 +812,8 @@ pyautogui.dragTo(tag#1, button='left') ``` When you think you can directly output precise x and y coordinates or there is no tag on which you want to interact, you can also use them directly. But you should be careful to ensure that the coordinates are correct. -Return one line or multiple lines of python code to perform the action each time, be time efficient. +Return one line or multiple lines of python code to perform the action each time, be time efficient. When predicting multiple lines of code, make some small sleep like `time.sleep(0.5);` interval so that the machine could take +You need to to specify the coordinates of by yourself based on your observation of current observation, but you should be careful to ensure that the coordinates are correct. You ONLY need to return the code inside a code block, like this: ```python # your code here @@ -816,6 +823,7 @@ When you think you have to wait for some time, return ```WAIT```; When you think the task can not be done, return ```FAIL```, don't easily say ```FAIL```, try your best to do the task; When you think the task is done, return ```DONE```. +My computer's password is 'password', feel free to use it when you need sudo rights. First give the current screenshot and previous things we did a short reflection, then RETURN ME THE CODE OR SPECIAL CODE I ASKED FOR. NEVER EVER RETURN ME ANYTHING ELSE. """.strip() @@ -845,7 +853,7 @@ Then, based on your analysis, in conjunction with human desktop using habits and """ ACTION_GROUNDING_PROMPT_SEEACT = """ -You are required to use `pyautogui` to perform the action, but don't use the `pyautogui.locateCenterOnScreen` function to locate the element you want to operate with since we have no image of the element you want to operate with. +You are required to use `pyautogui` to perform the action grounded to the observation, but DONOT use the `pyautogui.locateCenterOnScreen` function to locate the element you want to operate with since we have no image of the element you want to operate with. DONOT USE `pyautogui.screenshot()` to make screenshot. You can replace x, y in the code with the tag of the element you want to operate with. such as: ```python pyautogui.moveTo(tag#3) @@ -854,7 +862,8 @@ pyautogui.dragTo(tag#1, button='left') ``` When you think you can directly output precise x and y coordinates or there is no tag on which you want to interact, you can also use them directly. But you should be careful to ensure that the coordinates are correct. -Return one line or multiple lines of python code to perform the action each time, be time efficient. +Return one line or multiple lines of python code to perform the action each time, be time efficient. When predicting multiple lines of code, make some small sleep like `time.sleep(0.5);` interval so that the machine could take +You need to to specify the coordinates of by yourself based on your observation of current observation, but you should be careful to ensure that the coordinates are correct. You ONLY need to return the code inside a code block, like this: ```python # your code here @@ -864,5 +873,6 @@ When you think you have to wait for some time, return ```WAIT```; When you think the task can not be done, return ```FAIL```, don't easily say ```FAIL```, try your best to do the task; When you think the task is done, return ```DONE```. +My computer's password is 'password', feel free to use it when you need sudo rights. First give the current screenshot and previous things we did a short reflection, then RETURN ME THE CODE OR SPECIAL CODE I ASKED FOR. NEVER EVER RETURN ME ANYTHING ELSE. """