havikz/cua-benchmark-atlas

CUA Benchmark Atlas

Interactive Hugging Face Space app for visualizing run reliability from CUA artifacts.

Local run

cd hf-space/cua-leaderboard
pip install -r requirements.txt
python app.py

Set a custom runs folder if needed:

# PowerShell
$env:CUA_RUNS_DIR = "C:/path/to/data/runs"
python app.py

Or pull runs directly from your HF bucket:

# PowerShell
$env:CUA_HF_BUCKET_RUNS_URI = "hf://buckets/havikz/bucket/runs"
python app.py

Publish to HF Space

hf repo create havikz/cua-benchmark-atlas --type space --sdk gradio
hf upload-large-folder havikz/cua-benchmark-atlas ./

To make the Space live against your bucket, set CUA_HF_BUCKET_RUNS_URI as a Space variable, for example hf://buckets/havikz/bucket/runs.