Files
sci-gui-agent-benchmark/evaluation_examples/examples/multi_apps/337d318b-aa07-4f4f-b763-89d9a2dd013f.json
yuanmengqi e433f35c1f feat: standardize configuration fields across all evaluation examples
- Add `fixed_ip` field to all 369 JSON files in examples directory
  - Set to `true` for 8 files listed in google_chrome.json multi_apps
  - Set to `false` for remaining 361 files
- Add `possibility_of_env_change` field to 363 JSON files missing this field
  - Set to "low" for newly added fields
  - Preserve existing values (4 medium, 2 high) for 6 files that already had this field

This ensures consistent configuration schema across all evaluation examples
while maintaining backward compatibility with existing settings.
2025-07-16 13:45:34 +00:00

52 lines
2.1 KiB
JSON

{
"id": "337d318b-aa07-4f4f-b763-89d9a2dd013f",
"snapshot": "libreoffice_calc",
"instruction": "Cross-check the invoices with the bank statements and identify any discrepancies. Then pull out the invoices that don't match the statements and put them in the \"problematic\" folder.",
"source": "authors",
"config": [
{
"type": "download",
"parameters": {
"files": [
{
"path": "/home/user/Desktop/invoice TII-20220301-90.pdf",
"url": "https://huggingface.co/datasets/xlangai/ubuntu_osworld_file_cache/resolve/main/multi_apps/337d318b-aa07-4f4f-b763-89d9a2dd013f/invoice%20TII-20220301-90.pdf"
},
{
"path": "/home/user/Desktop/Invoice # GES-20220215-82.pdf",
"url": "https://huggingface.co/datasets/xlangai/ubuntu_osworld_file_cache/resolve/main/multi_apps/337d318b-aa07-4f4f-b763-89d9a2dd013f/Invoice%20%23%20GES-20220215-82.pdf"
},
{
"path": "/home/user/Desktop/Invoice # 243729.pdf",
"url": "https://huggingface.co/datasets/xlangai/ubuntu_osworld_file_cache/resolve/main/multi_apps/337d318b-aa07-4f4f-b763-89d9a2dd013f/Invoice%20%23%20243729.pdf"
},
{
"path": "/home/user/Desktop/Bank-Statement.pdf",
"url": "https://huggingface.co/datasets/xlangai/ubuntu_osworld_file_cache/resolve/main/multi_apps/337d318b-aa07-4f4f-b763-89d9a2dd013f/Bank-Statement.pdf"
}
]
}
}
],
"trajectory": "trajectories/337d318b-aa07-4f4f-b763-89d9a2dd013f",
"related_apps": [
"os",
"pdf"
],
"evaluator": {
"func": "compare_pdfs",
"result": {
"type": "vm_file",
"path": "/home/user/Desktop/problematic/Invoice # 243729.pdf",
"dest": "Invoice # 243729.pdf"
},
"expected": {
"type": "cloud_file",
"path": "https://huggingface.co/datasets/xlangai/ubuntu_osworld_file_cache/resolve/main/multi_apps/337d318b-aa07-4f4f-b763-89d9a2dd013f/Invoice%20%23%20243729.pdf",
"dest": "Invoice # 243729 Gold.pdf"
}
},
"proxy": false,
"fixed_ip": false,
"possibility_of_env_change": "low"
}