lzy/sci-gui-agent-benchmark
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Add hosted GBOX agent for OSWorld evaluation (#376)

Browse Source
This commit is contained in:
Subash Shibu
2025-11-12 21:13:31 -08:00
committed by GitHub
parent 00b6468eb7
commit 3167339e45
5 changed files with 736 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
monitor/.env
View File

@@ -2,13 +2,13 @@
# Do not write any secret keys or sensitive information here.
# Monitor configuration
TASK_CONFIG_PATH=../evaluation_examples/test_nogdrive.json
TASK_CONFIG_PATH=../evaluation_examples/test_50_random_proportional.json
EXAMPLES_BASE_PATH=../evaluation_examples/examples
RESULTS_BASE_PATH=../result_multi_apps_pengxiang_transformers12
# ACTION_SPACE=pyautogui
# OBSERVATION_TYPE=screenshot
# MODEL_NAME=computer-use-preview
# MAX_STEPS=150
FLASK_PORT=9001
RESULTS_BASE_PATH=../results_hosted_gbox_50
ACTION_SPACE=pyautogui
OBSERVATION_TYPE=screenshot
MODEL_NAME=us.anthropic.claude-sonnet-4-5-20250929-v1:0
MAX_STEPS=15
FLASK_PORT=8080
FLASK_HOST=0.0.0.0
FLASK_DEBUG=false