Skip to content

Distribute and run LLMs with a single file.

1.5 KB
Raw Blame
[project]
name = "llamafile-integration-tests"
version = "0.1.0"
requires-python = ">=3.10"
dependencies = [
    "pytest>=7.0.0",
    "requests>=2.28.0",
]

[tool.pytest.ini_options]
testpaths = ["tests"]
python_files = ["test_*.py"]
python_functions = ["test_*"]
log_cli = true
log_cli_level = "WARNING"
log_cli_format = "%(asctime)s [%(levelname)s] %(message)s"
log_cli_date_format = "%H:%M:%S"
markers = [
    "slow: marks tests as slow (deselect with '-m \"not slow\"')",
    "multimodal: requires a vision-capable model",
    "tool_calling: requires a tool-capable model",
    "thinking: requires a thinking model (QwQ, DeepSeek-R1, etc.)",
    "gpu: tests GPU acceleration",
    "cpu: tests CPU-only execution",
    "server: tests server mode",
    "tui: tests TUI mode",
    "cli: tests CLI mode",
    "combined: tests combined (TUI+server) mode",
    "help: tests --help output (no model required)",
    "ssl: tests HTTPS/TLS support (serving and model download)",
    "online: tests that need network access (deselect with '-m \"not online\"')",
    "sandbox: tests pledge()/SECCOMP sandboxing (enforcement checks need Linux)",
    "bare_executable: requires a bare llamafile binary with no embedded model (skip for pre-built .llamafile bundles)",
    "determinism: requires reproducible output at temperature=0 (unreliable on gpt-oss models, whose long forced reasoning trace and MXFP4 quantization give a threaded matmul rounding difference more opportunities to flip a near-tied token; see tests/run_integration_tests.sh)",
]