Skip to content

[T-104] Experiment configs EXP-01…EXP-07 (configs/experiments/*.yaml) validated in CI #206

Closed
Reet24-del/RoboProbe
#3

Description

@CodewithJha

Task T-104 · Milestone M10 — Evaluation · Size S · Priority P0 · Epic #10

Implements FR-EXP-001

Branch chore/t-104-experiment-configs-exp-01-exp — one task = one branch = one PR (doc 32 §6)

Summary

Write the seven experiment YAMLs exactly per doc 19 §4 (strategies, ablation variants as named strategy param sets, seeds, budgets, d=8 and noise×2 variants), each with pre-registered hypotheses and metrics, and a CI test validating them against the experiment schema and planner.

Context / motivation

Pre-registration (frozen hyper-parameters from T-061) protects against p-hacking (doc 19 §5, §8).

Scope

In scope

  • Ablation variants defined as strategy param overrides (no code forks).
  • configs/spaces/warehouse_v1_d8.yaml variant.

Implementation guidance

  • Files: configs/experiments/exp01_k2d_main.yaml, exp02_analytic.yaml, exp03_ablation.yaml, exp04_gazebo.yaml, exp05_profiles.yaml, exp06_gz_replay.yaml, exp07_scalability.yaml (doc 19 §4).
  • Each declares id, hypotheses (H1–H6), backend, space (+ version), sut_profiles, strategies (name + params override; ablations as named variants, e.g. probe_no_diversity: {strategy: S-PROBE, params: {lambda_div: 0}}), seeds (count + base seed), budget, batch_size, metrics (MET ids), primary_metric, alpha, notes.
  • Frozen S-PROBE hyper-parameters (T-061) referenced by hash; a CI test recomputes the hash and fails on mismatch (pre-registration integrity).
  • roboprobe experiment plan <file> --dry-run prints cells and estimated compute; paste the summary as a YAML comment.
  • Standards: respect the layering contracts (doc 16 §2) and conventions in doc 34; no hard-coded tunables — thresholds/defaults come from typed settings (doc 27); no magic numbers (ruff PLR2004); public APIs typed (mypy strict) and documented.

Configuration keys

From the configuration reference (doc 27). Keys marked (new) must be added to the typed settings models and appear in the generated reference (T-003c).

  • experiments.default_seeds
  • experiments.alpha
  • experiments.max_concurrent_campaigns

Acceptance criteria

  • All configs validate and plan; counts match doc 19 §4 table.
  • Pre-registration hash check in CI.
  • EXP-06 expressed as a replay plan referencing EXP-04 outputs (resolved at run time, documented).
  • Tests listed below added and green in CI; coverage of new code meets the gate in doc 18 §3.
  • Docs, config reference and CHANGELOG Unreleased updated where behaviour, keys or interfaces changed.
  • Definition of Done (doc 32 §3) satisfied.

Tests required

  • tests/integration/experiments/test_configs.py.

Dependencies

Blocked by: #108 (T-102a), #122 (T-061)

Blocks: #207 (T-105a), #208 (T-105b)

Parallelization notes

Files / paths owned by this task — avoid editing other paths; shared files (pyproject.toml entry points, cli/main.py, api/routers/__init__.py, router.tsx, docker/compose.yaml, ci.yml) are append-only:

  • configs/experiments/**
  • configs/spaces/warehouse_v1_d8.yaml

Independent.

Definition of Done

Per doc 32 §3 and AGENTS.md §5: lint, format, mypy strict and import-linter green; tests added and passing; docs updated; PR template completed with Closes #<this issue>; Conventional Commit messages referencing T-104 (doc 16 §6); commits authored only by the repository owner identity (AGENTS.md §4); CI green; reviewer checklist (doc 32 §5.1) satisfied.

References

Design docs: doc 19 §4 · doc 19 §5 · doc 19 §8

Requirements: FR-EXP-001 (traceability: doc 28)

Activity

  1. added
    type:choreMaintenance / scaffolding
    area:experimentsExperiment runner and EXP-xx protocols
    area:evalEvaluation campaigns and success criteria
    priority:P0Critical path / blocks others
    taskChild task of a milestone epic
    on Nov 19, 2026
  2. added a commit that references this issue on Nov 21, 2026
    c62404e
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    area:evalEvaluation campaigns and success criteriaarea:experimentsExperiment runner and EXP-xx protocolsmilestone:M10M10 — Evaluationpriority:P0Critical path / blocks otherssize:S<= 1 daytaskChild task of a milestone epictype:choreMaintenance / scaffolding

    Projects

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions