fix: recover PR 1931 ingest via sweep reuse

[Oseltamivir]

Recover the missing official ingest from source run 28214046079 (PR #1931). The push-to-main ingest run 28459613726 failed at reuse-ingest-artifacts validation on duplicate reran-eval rows; PR #1965 added auto-dedupe of reran eval artifacts, so reuse now validates.

[Oseltamivir]

/reuse-sweep-run 28214046079

[github-actions]

Thanks for the contribution! For vLLM & SGLang, please ensure that your recipes is similar to the official vLLM recipes and/or the SGLang cookbook

• https://docs.sglang.io/cookbook/intro
• https://recipes.vllm.ai/

If it is not, please create a PR first before we can merge your single node PR into the master branch. Let's ensure that the documentation is first class such that the entire ML community can benefit from your hard work! Thank you

• https://github.com/vllm-project/recipes
• https://github.com/sgl-project/sglang/tree/main/docs_new

PR authors are responsible for ensuring that after merging, all GitHub Action jobs fully pass. A lot of the time, failures are just flakes and simply re-running the failed jobs will fix it. If re-running failed jobs is attempted, PR authors are responsible for ensuring it passes. See GitHub's docs on re-running failed jobs: https://docs.github.com/en/actions/how-tos/manage-workflow-runs/re-run-workflows-and-jobs#re-running-failed-jobs-in-a-workflow

As a rule of thumb, generally, PR authors should request a review & get a PR approval from the respective companies' CODEOWNERS before requesting a review from core maintainers.

If additional help is needed, PR authors can reach out to core maintainers over Slack.

---

感谢你的贡献！对于 vLLM 与 SGLang，请确保你的 recipe 与官方 vLLM recipes 和/或 SGLang cookbook 保持一致

• https://docs.sglang.io/cookbook/intro
• https://recipes.vllm.ai/

如果不一致，请先创建一个 PR，之后我们才能将你的单节点 PR 合并到 master 分支。让我们确保文档保持一流水准，使整个 ML 社区都能从你的辛勤工作中受益！谢谢

• https://github.com/vllm-project/recipes
• https://github.com/sgl-project/sglang/tree/main/docs_new

PR 作者有责任确保合并后所有 GitHub Action 任务完全通过。 很多时候失败只是偶发抖动（flake），重新运行失败的任务即可解决。如果选择重新运行失败的任务，PR 作者有责任确保其最终通过。参见 GitHub 关于重新运行失败任务的文档：https://docs.github.com/en/actions/how-tos/manage-workflow-runs/re-run-workflows-and-jobs#re-running-failed-jobs-in-a-workflow

一般而言，PR 作者应先向相应公司的 CODEOWNERS 请求审阅并获得 PR 批准，然后再请求核心维护者审阅。

如需更多帮助，PR 作者可通过 Slack 联系核心维护者。

[Oseltamivir]

Recovery verified ✅

Official ingest for PR #1931 (minimaxm3-fp4-b300-dynamo-vllm) recovered via artifact reuse.

Failed target: run 28459613726 (push-to-main for PR #1931 merge cdb6db07) — reuse-ingest-artifacts failed on duplicate reran-eval rows, trigger-ingest skipped → no ingest.

Root cause: PR #1931 ran evals 6× due to flakes, leaving duplicate raw dirs + aggregate rows for the conc4096 config. Fixed by #1965 (auto-dedupe reran eval artifacts, keep latest result per identity).

Reused source: run 28214046079 attempt 6, SHA 8602631b (PR #1931 branch). Recovered config execution semantics verified unchanged since that SHA.

Recovery:

• merge commit cb90b3e0
• push-to-main recovery run 28485856738 — setup (reuse-enabled) ✓, reuse-ingest-artifacts ✓ (auto-deduped 9→7 eval rows, kept latest), upload-changelog-metadata ✓, trigger-ingest ✓
• downstream ingest 28485915878 ✓ — trigger RUN_ID 28485856738, reused source 28214046079 attempt 6

Ingested: eval results +7, eval samples +9233 across 7 files, benchmark results 30, run stats 9, changelog +1. DB verify, run overrides, and cache invalidation all succeeded; no unmapped entities. The conc4096/decode-tp8 config carries the latest rerun's accuracy (em_strict 0.9515), not the earlier flaky values.