Trust plan
Status for 0.1.0a13: software-side alpha complete; hardware validation pending.
Software gates completed
- Incomplete GuideLLM and vLLM request-count runs are ineligible for ranking.
- Explicit benchmark timeouts remain authoritative; duration-based limits are derived conservatively.
- Server drain evidence is required and persisted before a trial completes.
- Public synthetic tests cover configuration, execution, scheduling, retry, process ownership, crash finalization, redaction, scoring, reporting, and offline workflows.
- The failure percentage is an eligibility gate. The configured maximize metric is the primary objective everywhere; request failure rate and count are deterministic tie-breakers.
- Trusted defaults use one warm-up and three measured repeats. Reports use Student's t intervals, label fewer than three measurements exploratory, and state when drift cannot be evaluated.
- Coordinator failures atomically finalize recoverable runs as
failed; keyboard interruption finalizes them asinterrupted. - Credentials are recursively redacted from terminal output, manifests, run results, CSV, and HTML.
- CI owns Ruff, formatting, type, coverage, strict documentation, wheel installation, dependency audit, SBOM, package metadata, and tag/version gates on Python 3.11 and 3.12.
Hardware validation pending
No native-Linux, GPU, or real-backend result is claimed by this release. The following evidence must be produced on rented hardware before promotion beyond the software alpha:
- Native-Linux vLLM 0.28 end-to-end execution.
- L40 and H100 runs, including tensor parallel and at least one multi-GPU run.
- Real long-generation timeout and interruption cleanup.
- Raw GuideLLM versus
vllm bench servecomparison evidence using identical requests. - Execution of the repository GPU smoke workflow.
Record GPU, driver, CUDA, Python, vLLM, GuideLLM, model, configuration, raw artifacts, and dated
pass/fail outcomes in compatibility.md. Until those checks succeed, the project remains explicitly
hardware validation pending.