Commit the LDI local-decision architecture contract and expand the source inventory (SLM-114) - #331
Conversation
…entory (SLM-114) Make the local-decision-intervention (LDI) research synthesis and architectural boundaries canonical so future agents cannot recreate a parallel harness, repeat the falsified E249-E284 chain, or treat a local-metric gain as promotion evidence. - local-decision-interventions.md: add the LDI0-01 architecture contract (evidence/objective/actuator/experiment/promotion separation with named owners, six invariants, forbids a second stack, DecisionEventV2 as the next contract) and a measured-chain / current-blocker summary (blocker: stable state support does not imply objective/action-partition support) - local-decision-sources.json: add 9 arXiv works (MGDA, PCGrad, LoRA, PICARD, AdaLoRA, DeepSeekMath/GRPO, GAD/ASAp, min-p, TAB-PO) with metadata verified against arXiv; 33->42 rows, 25->34 papers, unique ids/URIs - research-lineage.md: extend the exact-state local-decision entry with the new sources under the existing Faithful/Adapted/Surrogate/Adjacent vocabulary - quality-experiment-matrix.md: add the LDI campaign index (namespace pointer; claims no unrun row, mints no E ID, records next-free >= E292) - test_harness.py: update+extend the manifest test (counts, source-id uniqueness, required metadata, unique canonical URIs, implementation-status vocabulary) - add iter-ldi0-01-source-expansion-20260717.md evidence memo Documentation/inventory only: no code, model, checkpoint, adapter, or model-card quality change; no ship gate weakened. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KVT9Qe5YUhgZqvaHLCAYSS
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
|
Warning Review limit reached
Next review available in: 46 minutes Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available. How can I continue?After more reviews become available, a review can be triggered using the To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews. How do review limits work?CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability. For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window. Please refer docs for additional details. Review details⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (1)
📝 WalkthroughWalkthroughThe change formalizes the LDI0-01 architecture contract, expands the local-decision source inventory, strengthens manifest validation, and updates experiment-matrix and research-lineage records with measured-chain and provenance details. ChangesLDI0-01 contract and measured boundary
Manifest expansion and validation
Research and campaign alignment
Estimated code review effort: 3 (Moderate) | ~20 minutes Possibly related PRs
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
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. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@tests/test_autoresearch/test_harness.py`:
- Line 745: Update the allowed_status set in the test harness to exactly match
the canonical architecture vocabulary: Faithful, Adapted, Surrogate, and
Adjacent. Remove Direct and Rejected while preserving the existing validation
flow that consumes allowed_status.
- Around line 753-754: Update the manifest read in the test harness to call
Path.read_text with an explicit UTF-8 encoding, matching _load_sources and
avoiding dependence on the host locale.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro Plus
Run ID: 873a59da-5b7c-4ff0-8ac0-416c0fb30680
📒 Files selected for processing (6)
docs/design/iter-ldi0-01-source-expansion-20260717.mddocs/design/local-decision-interventions.mddocs/design/quality-experiment-matrix.mddocs/design/research-lineage.mdsrc/slm_training/resources/autoresearch/local-decision-sources.jsontests/test_autoresearch/test_harness.py
…manifest UTF-8 (SLM-114)
- allowed_status now matches the repo's canonical research-lineage vocabulary
(Faithful/Adapted/Surrogate/Adjacent) rather than the issue text's Direct/Rejected;
the manifest values {Adapted, Adjacent} remain a subset, so the check is unchanged.
- read the manifest with explicit encoding="utf-8" to match _load_sources and avoid
host-locale dependence (author names contain non-ASCII characters).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KVT9Qe5YUhgZqvaHLCAYSS
2 participants
Summary
LDI0-01 (SLM-114). Makes the local-decision-intervention (LDI) research synthesis and architectural boundaries canonical in the repository so future agents cannot recreate a parallel harness, repeat the falsified E249-E284 chain, or treat a local-metric gain as promotion evidence. Documentation / research-inventory / test contract only — no code, model, checkpoint, adapter, or model-card quality change.
Changes
docs/design/local-decision-interventions.md— new Architecture contract (LDI0-01): evidence / objective / actuator / experiment / promotion separated into distinct named owners (local_decisions.py,trace_store.py,local_train.py, the quality matrix, ship gates + model card), six invariants, an explicit "do not create a second orchestration/training stack" rule, andDecisionEventV2as the next evidence contract. Added a Measured chain and current blocker section — the E248-E286 chain is negative and the current blocker is stated: stable state support does not imply objective/action-partition support.src/slm_training/resources/autoresearch/local-decision-sources.json— added 9 arXiv works (33→42 rows, 25→34 papers), each with metadata verified against arXiv on 2026-07-17: MGDA, Gradient Surgery/PCGrad, LoRA, PICARD, AdaLoRA, DeepSeekMath/GRPO, Grammar-Aligned Decoding/ASAp, min-p, TAB-PO. Uniquesource_ids and canonical URIs; no paper duplicated under an alternate URL.docs/design/research-lineage.md— extended the existing Exact-state local decision preference entry with the new sources using the repo's real Faithful / Adapted / Surrogate / Adjacent vocabulary (not duplicating LoRA/PICARD/GAD/GRPO entries that already exist).docs/design/quality-experiment-matrix.md— added the LDI campaign index: a namespace pointer that claims no unrun row and mints no E ID, records the negative measured chain, and codifies the E-ID allocation rule (highest allocated = E291; next free ≥ E292).tests/test_autoresearch/test_harness.py— updated and extendedtest_local_decision_source_manifest_is_completeto the new counts plus source-ID uniqueness, required-metadata presence, unique canonical URIs, implementation-status vocabulary, and the nine additions.docs/design/iter-ldi0-01-source-expansion-20260717.md— evidence memo (reviewed base commitc7cb099, before/after counts, duplicate/alternate-URL handling, classification table, honesty boundary).Acceptance criteria
local-decision-sources.jsonvalidates under_load_sourceswith no duplicate papers under alternate URLs (42 rows, unique ids + URIs).local-decision-interventions.mddistinguishes current implementation, measured evidence, proposed work, and the falsified chain.research-lineage.mdmaps each new derivative to a paper and labels it honestly.quality-experiment-matrix.mdhas an LDI namespace/index with no unrun-row claim and no conflicting E ID.ruffand the focused docs/source tests pass.Verification
Run against a Python 3.12
.venv(.[dev]):pytest tests/test_autoresearch -q→ 51 passed.githooks/check-changed→ selectstest_autoresearch/test_harness.py; ruff +py_compile+ pytest → 43 passedpython -m scripts.repo_policy→ ok ·git diff --check→ clean · all repository-relative links in the four changed docs resolveHonesty boundary
Specification / inventory only. No experiment, train, eval, benchmark, checkpoint, adapter, or model-card quality update; no paper result is represented as a repository result; constraint shadows certify decoder legality only, never a semantic preference label. No ship gate is weakened and no readiness or promotion claim is made.
🤖 Generated with Claude Code
https://claude.ai/code/session_01KVT9Qe5YUhgZqvaHLCAYSS
Generated by Claude Code
Summary by CodeRabbit
Documentation
Tests