Add latent-objective recognition eval to multi-objective skill

[cafzal]

Description

Adds a fifth eval to the cuopt-multi-objective-exploration skill — multiobj-explore-eval-005-latent-objective — covering the boundary the existing four don't: a problem stated with a single objective while a second objective sits latent in the data, unstated.

The current evals all hand the agent both objectives (001 interpret, 002 explore, 004 dual-as-slope) or are explicitly single-objective (003 decoy). None test recognizing a latent objective. This one grades whether the skill makes the agent surface the latent cost objective and trace the supply-vs-cost frontier — rather than optimizing the stated objective alone or silently folding cost into a self-chosen weighted blend (maximize supply − λ·cost). It brackets the skill's activation boundary opposite the 003 decoy.

Behavioral eval (expected_script: null, LLM-graded on the behavior list), same house style as 001/002/004; validate_skills.sh picks up the new array entry and the signature / BENCHMARK.md / skill-card regenerate via NVSkills-Eval. The latent-objective shape is the max-supply supply-vs-cost case validated on cuOpt (Tesla T4) in NVIDIA/cuopt-examples#157.

Checklist

• I am familiar with the Contributing Guidelines.
• Testing
  • New or existing tests cover these changes
  • Added tests
  • Created an issue to follow-up
  • NA
• Documentation
  • The documentation is up to date with these changes
  • Added new documentation
  • NA

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[coderabbitai]

📝 Walkthrough

Walkthrough

A single new evaluation case, multiobj-explore-eval-005-latent-objective, is appended to the evals array in skills/cuopt-multi-objective-exploration/evals/evals.json. The case defines a multi-period production planning problem with unstated cost objectives and specifies expected skill behavior for latent-objective discovery via epsilon-constraint Pareto tracing.

Changes

Latent Objective Eval Case

Layer / File(s)

Summary

New latent-objective evaluation case skills/cuopt-multi-objective-exploration/evals/evals.json

Adds multiobj-explore-eval-005-latent-objective with a question describing a production planning scenario where cost data is present but not in the stated objective. Specifies expected outputs: epsilon-constraint Pareto frontier tracing, exchange-rate estimation via adjacent-point differencing (MILP/no duals), interpretable operating points with knee flagging, and exclusion of single-plan collapse or self-chosen weighted-sum behaviors.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~3 minutes

Possibly related PRs

• NVIDIA/cuopt#1406: Updates cuopt-multi-objective-exploration evaluation expectations and documentation on exchange-rate derivation (dual when available, otherwise differencing), directly matching the method specified in the new eval case.

Suggested labels

non-breaking, improvement

Suggested reviewers

• mlubin
• rgsl888prabhu
• tmckayus

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly and specifically identifies the main change: adding a new evaluation case for latent-objective recognition to the multi-objective exploration skill.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Description check	✅ Passed	The PR description clearly explains the new evaluation case, its purpose, and how it complements existing evaluations by testing latent objective recognition.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[cafzal]

@ramakrishnap-nv small skill eval – adds the latent-objective case (the under-trigger boundary the other four miss) to cuopt-multi-objective-exploration, mirroring the existing four. Ready when you/Miles have a cycle.

[ramakrishnap-nv]

/ok to test 8275bf7

[ramakrishnap-nv]

/merge