Skip to content

armgit / cua

Public Imported from GitHub
0

Scale computer-use 2.0 with open-source drivers, cross-OS sandboxes, and benchmarks for training, evaluation, and data generation.

703 B
Raw Blame
# One task per line. `check` runs in the task's directory inside the sandbox
# (exit 0 passes). `mock` scripts the mock provider (cua-mock-llm) and is
# used only with --scripted.
{"id": "hello", "prompt": "Create a file hello.txt whose only line is: hello", "check": "grep -qx hello hello.txt", "mock": "shell echo hello > hello.txt && cat hello.txt"}
{"id": "primes", "prompt": "Write primes.py that prints the first 10 primes, one per line, and run it.", "check": "test \"$(python3 primes.py | tr '\\n' ' ')\" = '2 3 5 7 11 13 17 19 23 29 '", "mock": "shell printf '%s\\n' 'p = [n for n in range(2, 30) if all(n % d for d in range(2, n))]' 'print(*p, sep=chr(10))' > primes.py && python3 primes.py"}