spb/wpai0001_uqo

1# Implementation task tracker — WP-AI-000123Legend: [x] done · [~] in progress · [ ] todo · (gate) quality gate per CLAUDE.md §1045## Sprint 0 — Research audit6- [x] Locate UQO Research workspace, conventions, remotes7- [x] Environment audit (laptop + cluster) → `docs/ENVIRONMENT_AUDIT.json`8- [x] Literature verification → `literature/`9- [x] RESEARCH_GAP, PREREGISTRATION v0.1, EXPERIMENT_PLAN, COMPUTE_BUDGET, RISK_REGISTER, REPRODUCIBILITY, logs10- [x] Project skeleton, pyproject, tests (Gate A)11- [x] `scripts/build_figures.py`, `scripts/build_paper.sh`, paper skeleton compiles (placeholders)12- [x] git repo + `origin = gitsrv:srv/git/wpai0001_uqo.git`1314## Sprint 1 — Controlled micro-world15- [x] OPSEQ / FSA / MORPH generators + exact validators + disjointness + hashing16- [x] Toy transformer with hook points; no-op tests (Gate C part 1)17- [x] Training loop (teacher-forced scratchpad), greedy decoding, metrics, cluster bootstrap18- [x] Pilot run (3 seeds) — gap qualification (Gate B passed; run `pilot_tier0_opseq_451ef5087182`)19- [ ] FSA / MORPH toy baselines20- [ ] `trace_stride = 2` difficulty variant baseline2122## Sprint 2 — Causal discovery23- [x] Within-donor localization (patching, ablation, heads) with sham baseline and reproducibility (RQ1 pilot: Ω executed at the read position; L0 attention + L0–L2 MLPs)24- [ ] Add `embed`/`resid_pre` family and path patching Ω-token → r1 to the sweep25- [ ] Subspace localization (DAS-style / PCA of Ω-contrast) with norm-matched random subspaces26- [ ] Necessity test on donor (H3) with control-stratum retention2728## Sprint 3 — First transplant (done 2026-11-09)29- [x] Alignment module: identity, Procrustes, ridge, CCA, random orthogonal; task-row vs task-free fits; leakage audit by item hash30- [x] Activation graft + sham + norm-matched random + shuffled-token/layer + unrelated-donor controls31- [x] Gate D: reproducibility across 9 pairs, overhead disclosed (donor in the loop), map parameter counts3233## Sprint 4–6 (done 2026-11-09)34- [x] Subspace, steering, adapter, weight, persistent grafts; supervised-adapter and distillation references35- [x] Layer grid, dose, width/depth mismatch, stride-2; ablation → rescue; reverse graft; rule-vs-cache36- [x] Confirmatory freeze (PREREGISTRATION v1.0, D10) and confirmatory run + analysis (H2, H5 supported; H4 not)37- [x] Tier-1 MLX layer wrapping with bitwise no-op test; task-vector localization + transfer script38- [~] Tier-1 development run (4 pairs) → Tier-1 freeze → confirmatory split run39- [~] Tier-2: Qwen2.5-7B ↔ 3B pairs on M3U96b40- [ ] Transformers equivalence test for the MLX backend (not done; reported as a limitation)4142## Sprint 7 — Manuscript43- [~] Paper sections written for Tier 0; Tier 1 pending; claim-evidence table (Gate F) pending44- [ ] Internal multi-agent review (UQO-WP-REVIEW pipeline)45