Skip to content

fix(benchmark): require declared workflow depth and token budgets - #1091

Draft
seonghobae wants to merge 3 commits into
codex/declared-paired-bootstrap-20260907from
codex/declared-workflow-budget-20260907
Draft

fix(benchmark): require declared workflow depth and token budgets#1091
seonghobae wants to merge 3 commits into
codex/declared-paired-bootstrap-20260907from
codex/declared-workflow-budget-20260907

Conversation

@seonghobae

Copy link
Copy Markdown
Contributor

Stack

Draft successor of #1090 (5e952933). Parent #1074 and psychometric KPI owner #1067 stay open. This PR does not close #1090.

Exact head: 6a32c676eada0e3b49b98f5093a0cf3c13b7b981.

Problem

NIM equal-budget cells used a hidden five-step workflow and a 264-token per-call output cap. Those numbers allocated evaluation compute and shaped request planning without an operator declaration.

Change

  • maximum_calls / max_workflow_depth and max_output_tokens / total_token_budget are required positive integers. None is a fail-closed sentinel, not a default.
  • The equal cell token budget is the product of the two declarations.
  • CLI --max-workflow-depth and --max-output-tokens have no hidden default. Workflow YAML still writes 5 and 264 as this run's choices.
  • Report schema stays 4.0.0; those provenance fields already existed and are now validated as declarations.
  • ADR 0043 is Proposed. Production route/conduct defaults stay locked.

Verification

Local three-file coverage command matching hosted quality:

  • 181 passed in 3.71s
  • nim_benchmark.py 1331 statements / 502 branches, 0 missed
  • interrogate 100%

This is calculation-contract evidence, not buyer-held-out accuracy, p95 latency, or protected merge.

Remaining no-heuristics work

The psychometric held-out harness's 2,000-sample 95% interval on parent #1067 remains a later slice. No production policy or release is authorized.

RED: omitted maximum_calls, max_output_tokens, and max_workflow_depth
still succeed through hidden module defaults. The next commit must fail
closed without inventing a five-step envelope or 264-token cap.
Remove hidden five-step and 264-token defaults from request planning,
equal-budget cells, CLI, and provenance. Missing or non-positive
declarations fail closed. Workflow YAML still writes 5 and 264 as this
run's choices. Production route/conduct defaults stay locked.
ADR 0043 is Proposed. Gap baseline, changelog, doctoring, and NIM
operator docs keep 5 and 264 as run declarations. Held-out 2,000-sample
interval remains later work.
@coderabbitai

coderabbitai Bot commented Sep 7, 2026

Copy link
Copy Markdown

Important

Draft PR not reviewed

Draft PRs are not automatically reviewed by default.

  • Trigger a manual review

To automatically review draft PRs, update your CodeRabbit configuration:

reviews:
  auto_review:
    drafts: true

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

Comment @coderabbitai help to get the list of available commands.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

bug Something isn't working priority: medium Normal-priority or P2 work

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant