lzy/sci-gui-agent-benchmark
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: 本地修改 agent.py / run_proxmox / chrome tasks

Browse Source
This commit is contained in:
lizhanyuan
2026-03-19 14:49:39 +08:00
parent fe1bdae9a6
commit 970d430dcf
16 changed files with 707 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

45
evaluation_examples/examples/chrome_windows/chrome_win_task01.json Normal file
View File

@@ -0,0 +1,45 @@
{
"id": "chrome_win_task01",
"snapshot": "snapshot",
"instruction": "Can you make a new folder for me on the bookmarks bar in Chrome? Let's call it 'Favorites.'",
"source": "OSWorld-adapted",
"config": [
{
"type": "launch",
"parameters": {
"command": [
"C:\\Program Files\\Google\\Chrome\\Application\\chrome.exe"
],
"shell": false
}
},
{
"type": "sleep",
"parameters": {
"seconds": 5
}
}
],
"trajectory": "trajectories/",
"related_apps": [
"chrome"
],
"evaluator": {
"postconfig": [
{
"type": "sleep",
"parameters": {
"seconds": 3
}
}
],
"func": "vllm_eval"
},
"proxy": false,
"fixed_ip": true,
"possibility_of_env_change": "low",
"metadata": {
"input_files": [],
"steps": "1. Chrome已自动打开。\n2. 在书签栏空白处右键，选择'Add folder'（或'添加文件夹'）。\n3. 在弹出的对话框中将文件夹名称设为'Favorites'，点击保存。\n4. 确认书签栏上出现了名为'Favorites'的文件夹。"
}
}

46
evaluation_examples/examples/chrome_windows/chrome_win_task02.json Normal file
View File

@@ -0,0 +1,46 @@
{
"id": "chrome_win_task02",
"snapshot": "snapshot",
"instruction": "Can you save this webpage I'm looking at to the bookmarks bar so I can come back to it later?",
"source": "OSWorld-adapted",
"config": [
{
"type": "launch",
"parameters": {
"command": [
"C:\\Program Files\\Google\\Chrome\\Application\\chrome.exe",
"https://jalammar.github.io/illustrated-transformer/"
],
"shell": false
}
},
{
"type": "sleep",
"parameters": {
"seconds": 5
}
}
],
"trajectory": "trajectories/",
"related_apps": [
"chrome"
],
"evaluator": {
"postconfig": [
{
"type": "sleep",
"parameters": {
"seconds": 3
}
}
],
"func": "vllm_eval"
},
"proxy": false,
"fixed_ip": true,
"possibility_of_env_change": "low",
"metadata": {
"input_files": [],
"steps": "1. Chrome已自动打开并导航到目标页面。\n2. 按 Ctrl+D 打开书签保存对话框，或点击地址栏右侧的星形图标。\n3. 在'Folder'下拉菜单中选择'Bookmarks bar'（书签栏）。\n4. 点击'Done'保存。\n5. 确认书签栏上出现了该页面的书签。"
}
}

45
evaluation_examples/examples/chrome_windows/chrome_win_task03.json Normal file
View File

@@ -0,0 +1,45 @@
{
"id": "chrome_win_task03",
"snapshot": "snapshot",
"instruction": "My grandmother has been using Chrome and told me the font size is way too small. Could you set the default font size to the largest?",
"source": "OSWorld-adapted",
"config": [
{
"type": "launch",
"parameters": {
"command": [
"C:\\Program Files\\Google\\Chrome\\Application\\chrome.exe"
],
"shell": false
}
},
{
"type": "sleep",
"parameters": {
"seconds": 5
}
}
],
"trajectory": "trajectories/",
"related_apps": [
"chrome"
],
"evaluator": {
"postconfig": [
{
"type": "sleep",
"parameters": {
"seconds": 3
}
}
],
"func": "vllm_eval"
},
"proxy": false,
"fixed_ip": true,
"possibility_of_env_change": "low",
"metadata": {
"input_files": [],
"steps": "1. Chrome已自动打开。\n2. 点击右上角三点菜单 → Settings（设置）。\n3. 在左侧找到'Appearance'（外观），点击进入。\n4. 找到'Font size'（字体大小），将其设置为'Very large'（最大）。\n5. 确认字体大小选项显示为'Very large'。"
}
}

45
evaluation_examples/examples/chrome_windows/chrome_win_task04.json Normal file
View File

@@ -0,0 +1,45 @@
{
"id": "chrome_win_task04",
"snapshot": "snapshot",
"instruction": "Can you make Bing the main search engine when I look stuff up in Chrome?",
"source": "OSWorld-adapted",
"config": [
{
"type": "launch",
"parameters": {
"command": [
"C:\\Program Files\\Google\\Chrome\\Application\\chrome.exe"
],
"shell": false
}
},
{
"type": "sleep",
"parameters": {
"seconds": 5
}
}
],
"trajectory": "trajectories/",
"related_apps": [
"chrome"
],
"evaluator": {
"postconfig": [
{
"type": "sleep",
"parameters": {
"seconds": 3
}
}
],
"func": "vllm_eval"
},
"proxy": false,
"fixed_ip": true,
"possibility_of_env_change": "low",
"metadata": {
"input_files": [],
"steps": "1. Chrome已自动打开。\n2. 点击右上角三点菜单 → Settings（设置）。\n3. 在左侧点击'Search engine'（搜索引擎）。\n4. 在'Search engine used in the address bar'下拉菜单中选择'Bing'。\n5. 确认搜索引擎已改为Bing。"
}
}

45
evaluation_examples/examples/chrome_windows/chrome_win_task05.json Normal file
View File

@@ -0,0 +1,45 @@
{
"id": "chrome_win_task05",
"snapshot": "snapshot",
"instruction": "I want Chrome to enable the 'Do Not Track' feature to enhance my online privacy. Can you turn that on?",
"source": "OSWorld-adapted",
"config": [
{
"type": "launch",
"parameters": {
"command": [
"C:\\Program Files\\Google\\Chrome\\Application\\chrome.exe"
],
"shell": false
}
},
{
"type": "sleep",
"parameters": {
"seconds": 5
}
}
],
"trajectory": "trajectories/",
"related_apps": [
"chrome"
],
"evaluator": {
"postconfig": [
{
"type": "sleep",
"parameters": {
"seconds": 3
}
}
],
"func": "vllm_eval"
},
"proxy": false,
"fixed_ip": true,
"possibility_of_env_change": "low",
"metadata": {
"input_files": [],
"steps": "1. Chrome已自动打开。\n2. 点击右上角三点菜单 → Settings（设置）。\n3. 在左侧点击'Privacy and security'（隐私和安全）。\n4. 找到'Send a Do Not Track request with your browsing traffic'，将其开关打开（变为蓝色）。\n5. 确认该开关处于开启状态。"
}
}

45
evaluation_examples/examples/chrome_windows/chrome_win_task06.json Normal file
View File

@@ -0,0 +1,45 @@
{
"id": "chrome_win_task06",
"snapshot": "snapshot",
"instruction": "Please help me set Chrome to delete my browsing data automatically every time I close the browser.",
"source": "OSWorld-adapted",
"config": [
{
"type": "launch",
"parameters": {
"command": [
"C:\\Program Files\\Google\\Chrome\\Application\\chrome.exe"
],
"shell": false
}
},
{
"type": "sleep",
"parameters": {
"seconds": 5
}
}
],
"trajectory": "trajectories/",
"related_apps": [
"chrome"
],
"evaluator": {
"postconfig": [
{
"type": "sleep",
"parameters": {
"seconds": 3
}
}
],
"func": "vllm_eval"
},
"proxy": false,
"fixed_ip": true,
"possibility_of_env_change": "low",
"metadata": {
"input_files": [],
"steps": "1. Chrome已自动打开。\n2. 点击右上角三点菜单 → Settings（设置）。\n3. 点击'Privacy and security' → 'Cookies and other site data'。\n4. 找到'Clear cookies and site data when you close all windows'，将其开关打开。\n5. 确认该开关处于开启状态。"
}
}

45
evaluation_examples/examples/chrome_windows/chrome_win_task07.json Normal file
View File

@@ -0,0 +1,45 @@
{
"id": "chrome_win_task07",
"snapshot": "snapshot",
"instruction": "I want to update my Chrome profile name to Thomas. Could you help me change the username in Chrome profiles?",
"source": "OSWorld-adapted",
"config": [
{
"type": "launch",
"parameters": {
"command": [
"C:\\Program Files\\Google\\Chrome\\Application\\chrome.exe"
],
"shell": false
}
},
{
"type": "sleep",
"parameters": {
"seconds": 5
}
}
],
"trajectory": "trajectories/",
"related_apps": [
"chrome"
],
"evaluator": {
"postconfig": [
{
"type": "sleep",
"parameters": {
"seconds": 3
}
}
],
"func": "vllm_eval"
},
"proxy": false,
"fixed_ip": true,
"possibility_of_env_change": "low",
"metadata": {
"input_files": [],
"steps": "1. Chrome已自动打开。\n2. 点击右上角的个人头像图标（Profile图标）。\n3. 在弹出菜单中点击铅笔编辑图标或'Manage profiles'。\n4. 在名称输入框中将当前名称修改为'Thomas'。\n5. 点击完成/保存，确认界面上显示用户名为'Thomas'。"
}
}

68
evaluation_examples/examples/chrome_windows/chrome_win_task08.json Normal file
View File

@@ -0,0 +1,68 @@
{
"id": "chrome_win_task08",
"snapshot": "snapshot",
"instruction": "Can you make my computer bring back the last tab I shut down in Chrome?",
"source": "OSWorld-adapted",
"config": [
{
"type": "launch",
"parameters": {
"command": [
"C:\\Program Files\\Google\\Chrome\\Application\\chrome.exe",
"https://www.lonelyplanet.com"
],
"shell": false
}
},
{
"type": "launch",
"parameters": {
"command": [
"C:\\Program Files\\Google\\Chrome\\Application\\chrome.exe",
"--new-tab",
"https://www.airbnb.com"
],
"shell": false
}
},
{
"type": "launch",
"parameters": {
"command": [
"C:\\Program Files\\Google\\Chrome\\Application\\chrome.exe",
"--new-tab",
"https://www.tripadvisor.com"
],
"shell": false
}
},
{
"type": "sleep",
"parameters": {
"seconds": 5
}
}
],
"trajectory": "trajectories/",
"related_apps": [
"chrome"
],
"evaluator": {
"postconfig": [
{
"type": "sleep",
"parameters": {
"seconds": 3
}
}
],
"func": "vllm_eval"
},
"proxy": false,
"fixed_ip": true,
"possibility_of_env_change": "low",
"metadata": {
"input_files": [],
"steps": "1. Chrome已自动打开并加载了多个标签页。\n2. 关闭最右侧的标签页（点击该标签页上的X，或按Ctrl+W）。\n3. 恢复刚刚关闭的标签页：按 Ctrl+Shift+T，或右键标签栏选择'Reopen closed tab'。\n4. 确认被关闭的标签页重新出现在标签栏中。"
}
}

45
evaluation_examples/examples/chrome_windows/chrome_win_task09.json Normal file
View File

@@ -0,0 +1,45 @@
{
"id": "chrome_win_task09",
"snapshot": "snapshot",
"instruction": "Computer, please navigate to the area in Chrome settings where my passwords are stored.",
"source": "OSWorld-adapted",
"config": [
{
"type": "launch",
"parameters": {
"command": [
"C:\\Program Files\\Google\\Chrome\\Application\\chrome.exe"
],
"shell": false
}
},
{
"type": "sleep",
"parameters": {
"seconds": 5
}
}
],
"trajectory": "trajectories/",
"related_apps": [
"chrome"
],
"evaluator": {
"postconfig": [
{
"type": "sleep",
"parameters": {
"seconds": 3
}
}
],
"func": "vllm_eval"
},
"proxy": false,
"fixed_ip": true,
"possibility_of_env_change": "low",
"metadata": {
"input_files": [],
"steps": "1. Chrome已自动打开。\n2. 在地址栏输入 chrome://password-manager/passwords 并按回车，或通过 Settings → Autofill and passwords → Google Password Manager 找到密码管理页面。\n3. 确认页面显示的是Chrome密码管理器界面。"
}
}

46
evaluation_examples/examples/chrome_windows/chrome_win_task10.json Normal file
View File

@@ -0,0 +1,46 @@
{
"id": "chrome_win_task10",
"snapshot": "snapshot",
"instruction": "Please help me find the score record for the 2019 Super Bowl on the NFL website.",
"source": "OSWorld-adapted",
"config": [
{
"type": "launch",
"parameters": {
"command": [
"C:\\Program Files\\Google\\Chrome\\Application\\chrome.exe",
"https://www.nfl.com/"
],
"shell": false
}
},
{
"type": "sleep",
"parameters": {
"seconds": 5
}
}
],
"trajectory": "trajectories/",
"related_apps": [
"chrome"
],
"evaluator": {
"postconfig": [
{
"type": "sleep",
"parameters": {
"seconds": 3
}
}
],
"func": "vllm_eval"
},
"proxy": false,
"fixed_ip": true,
"possibility_of_env_change": "low",
"metadata": {
"input_files": [],
"steps": "1. Chrome已自动打开并导航到NFL官网。\n2. 在页面导航中找到Scores或Schedule入口。\n3. 找到2019赛季的Super Bowl比赛（Super Bowl LIII，2019年2月3日）。\n4. 确认页面上显示了该场比赛的比分信息。"
}
}

46
evaluation_examples/examples/chrome_windows/chrome_win_task11.json Normal file
View File

@@ -0,0 +1,46 @@
{
"id": "chrome_win_task11",
"snapshot": "snapshot",
"instruction": "Open the baggage fee calculator on the United Airlines website.",
"source": "OSWorld-adapted",
"config": [
{
"type": "launch",
"parameters": {
"command": [
"C:\\Program Files\\Google\\Chrome\\Application\\chrome.exe",
"https://www.united.com/en/us"
],
"shell": false
}
},
{
"type": "sleep",
"parameters": {
"seconds": 5
}
}
],
"trajectory": "trajectories/",
"related_apps": [
"chrome"
],
"evaluator": {
"postconfig": [
{
"type": "sleep",
"parameters": {
"seconds": 3
}
}
],
"func": "vllm_eval"
},
"proxy": false,
"fixed_ip": true,
"possibility_of_env_change": "low",
"metadata": {
"input_files": [],
"steps": "1. Chrome已自动打开并导航到United Airlines官网。\n2. 在页面导航中找到Travel info或Baggage部分。\n3. 找到并打开行李费计算器（Baggage Calculator）页面。\n4. 确认页面显示的是行李费计算器界面。"
}
}

46
evaluation_examples/examples/chrome_windows/chrome_win_task12.json Normal file
View File

@@ -0,0 +1,46 @@
{
"id": "chrome_win_task12",
"snapshot": "snapshot",
"instruction": "Show me the side effects of Tamiflu on the Drugs.com website.",
"source": "OSWorld-adapted",
"config": [
{
"type": "launch",
"parameters": {
"command": [
"C:\\Program Files\\Google\\Chrome\\Application\\chrome.exe",
"https://www.drugs.com/tamiflu.html"
],
"shell": false
}
},
{
"type": "sleep",
"parameters": {
"seconds": 5
}
}
],
"trajectory": "trajectories/",
"related_apps": [
"chrome"
],
"evaluator": {
"postconfig": [
{
"type": "sleep",
"parameters": {
"seconds": 3
}
}
],
"func": "vllm_eval"
},
"proxy": false,
"fixed_ip": true,
"possibility_of_env_change": "low",
"metadata": {
"input_files": [],
"steps": "1. Chrome已自动打开并导航到Drugs.com的Tamiflu页面。\n2. 在页面上找到'Side Effects'（副作用）部分并点击查看。\n3. 确认页面上显示了Tamiflu的副作用信息列表。"
}
}