diff --git a/README.md b/README.md index e1d5a389..40d9ed8d 100644 --- a/README.md +++ b/README.md @@ -92,9 +92,8 @@ tool for every SOLAR use. The architecture audit is packaged and content-addressed, with its throttled resource measurements explicitly limited to instruction/runtime corroboration. The reproducibly built mapper digest is reviewed and present in the release allowlist, so the formal producer reports -ready. Official scoring remains unavailable because the corpus policy is still -pending and no repository release bundle has been published. An executable -whose digest is only self-declared by its local provenance manifest is still +ready. The corpus scoring policy is available and the gfx1200 release bundle +is published under [`RELEASE/`](RELEASE/README.md). An executable whose digest is only self-declared by its local provenance manifest is still rejected: ```bash @@ -146,12 +145,14 @@ the default executor calls public PyTorch APIs independently of the ATen path. ## Official score -The schema v7 corpus pins the pending -`rx9060xt-gfx1200-reference-v2` baseline identity. Official scoring remains -unauthorized until its v2 release evidence is published. Once authorized, the -fail-closed scorer accepts a publisher-authored release bundle whose SHA-256 -references bind the corpus, baseline, candidate execution, and pinned SOLAR -manifests: +The schema v7 corpus pins the `rx9060xt-gfx1200-reference-v2` baseline +identity, and its v2 release evidence is published. The published gfx1200 +official SOL score is `0.497818327` over 43 problems and 163 workloads. Its +candidate is the eager PyTorch reference scored against itself, so the score +validates the scoring pipeline rather than any optimized kernel; see +[`RELEASE/README.md`](RELEASE/README.md). The fail-closed scorer accepts a +publisher-authored release bundle whose SHA-256 references bind the corpus, +baseline, candidate execution, and pinned SOLAR manifests: ```bash uv run sol-execbench score official RELEASE/release-bundle.json