Captions describe the data and state the finding, not the figure's role in
the paper. Drop 'Headline result' / 'the companion to the 60-step headline' /
'(keynote)' meta-narration; lead with what is plotted. Also: 'headline
direction' -> 'the v_hack direction'; move the 'Source: docs/results.md'
provenance from body text into a comment.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Addresses the formatting review:
- Figure 1 (keynote) moved to page 1 (declared before body, inline float)
- placeholder Introduction prose + hypothesis block (from README), \TODO rewrite
- direction arrows on every metric column (hack down-arrow, solve up-arrow);
best cells bold
- pseudocode -> algorithm/algpseudocode (math, not monospace ASCII); real
Python and the chat prompt stay lstlisting
- math/underscore removed from headings; loophole-mode names in code font
- ablation Source column moved into a comment (internal, not shown)
- long-run fig caption made explicitly the 200-step companion to the headline
- every float now has a text reference (placeholder where prose is TODO)
- dropped the 'honest (clean)' tic; added Q comment on the PackNet/LoRA bullet
(is it load-bearing or reviewer-driven?); TODO for a per-pairset example appendix
Builds clean: 11 pages, no unresolved refs/cites.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Formatting pass lifted from the AntiPaSTO paper (the format the author is
happy with):
- verbatim -> lstlisting (framed, shaded, Python-highlighted code blocks;
chat-template prompt uses language={} so markup isn't keyword-coloured)
- xcolor[table] + \rowcolor highlight on the 'ours' rows (keynote, ablation)
- ablation table restructured as leave-one-out with the negate symbol
(negate-routing/directional/hack-pairs/intervention); long interpretation
moved out of the caption into section body; post-hoc split into its own block
- real AntiPaSTO citation (Clark 2026, arXiv:2601.07473) replacing the
UNVERIFIED placeholder; dropped the verify-before-submission TODO
- code-availability line with a GitHub glyph (anonymous placeholder)
Builds clean: 11 pages, no unresolved refs/cites.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Move per-project agent memory into the repo so it is version-controlled.
Harness path /root/.claude/projects/.../memory now symlinks here, so
auto-load still works while the files live under git. Un-ignore only
.claude/memory/ (worktrees, locks, local settings stay ignored).
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Fill the \TODO{interp} in tab:ablation caption: post-hoc erase exposes how weak v_hack is
(weight-erase 0.39->0.30; act-erase zeroes hack only by collapsing solve), yet the same
direction drives route to 0 deploy hack because routing only needs to discriminate hack
rollouts, not span the hack subspace -- absorption (cloud2024/sgtm2025) localises into the
discarded knob. 'A detector too weak to erase a trained hack is still strong enough to route
one as it forms.'
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
The post-hoc erase result (weight 0.391->0.297) shows the rank-~10 v_hack is too weak to
span/erase the trained hack in W. But the same vector works at train time because a gate only
needs to DISCRIMINATE hack rollouts, and SGTM's absorption (Cloud 2024/2025) + self-reinforcing
localization amplify a weak noisy direction into full localization in the throwaway knob. This
is the mechanism A5 (held-out modes) tests -- logged as hypothesis, not yet shown in our RL setup.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Job 98 (tt_erase_bench on 20260531 vanilla ckpt, n=192): post-hoc erasure cannot
isolate the hack. weight_erase dents hack 0.391->0.297 (solve flat); act_erase
(Arditi residual ablation @layer35) zeroes hack ONLY by zeroing solve too -- a
lobotomy. Contrast: train-time route gets hack 0.000 AND solve 0.625. Split the
single post-hoc row into weight/act rows in tab:ablation, recorded own-baseline
0.391 in provenance. Journal 2026-06-03(c).
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Job 96 (erase static, frozen v_hack, s41) finished: deploy hack 0.500 / solve 0.500
(HACK_S 0.518). Both erase arms now in tab:ablation and both fail to suppress
(static 0.500, online 0.562) vs vanilla 0.359 and route 0.000. Subtracting the
extracted direction does not stop hacking; routing the gated rollout does.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Job 76 (erase online refresh-5 s41) finished: deploy hack 0.562 / solve 0.438.
One-sided gradient erasure ends ABOVE vanilla (0.359) at deploy -- it does not
suppress hacking, while route zeroes it. cos_post pinned 0 each step (we did
remove the aligned component) yet hack still emerged, so the hack signal lives
largely off the extracted axis under erase. Filled tab:ablation vanilla(77)+
erase-online(76) rows, corrected stale job-id mapping (96/86/87/88 after requeue).
Journal 2026-06-03(b).
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Header reprint fixes the variable-width misread trap (20+ unlabeled cols, gn
adjacent to lr). Records the anticipated Piggyback 'why not learn the routing
mask' critique (answer: no-cheat withholds the per-rollout label a learned mask
needs) and LoRA rank-deficiency as mild support for the low-rank hack subspace.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Tests the post-saturation collapse mechanism for vanilla long runs: as a
loophole saturates, every rollout in a group hacks -> identical reward ->
group hits the Dr.GRPO zero-variance skip -> no learning signal. Prediction:
zerovar climbs toward max right as lp_s starts collapsing (~step 80 in job 85).
Surfaced on the existing per-step diag debug line, not the streaming table.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Job 77 (vanilla s41) landed -> both arms n=3. Fill tab:keynote + fig:keynote
caption, add paired t-test, pin the exact 6-log regen command (just dyn
--latest-per-arm clobbers the band). Regenerated dyn_sub4 figure from the 6
explicit seed logs, fixing the 87cca9a clobber. Journal entry 2026-06-03(a).
Also: README points to main.tex and drops the stale n=1 findings block; record
two OpenReview URLs as a TODO in related work (mine reviews for shared critiques).
Closes A1/A2 (#173).
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Audit of all 4 plot scripts (plot_dynamics/substrate/emergence/deploy_overlay):
- One save_fig(fig, path) helper in figs.py writes png+svg+pdf (vector for the
paper, png for the blog). All scripts call it.
- arm_label() map: reader-facing names only -- route2->route, drop 'knob'/'the
cheat' from titles and the train-vs-deploy story (adapter on/off, reward hack).
- Titles off by default (the paper/blog caption carries it); --title re-enables
for standalone research use.
- dump_data CSV now carries every plotted series; plot_dynamics --from-csv
re-renders the three figures from the committed CSV with no logs (logs/ and
out/runs/ are gitignored; out/figs/*.csv is tracked). Round-trip verified.
- Commit the regenerated dyn_sub4 figures in all 3 formats + the CSV.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
- Inline author-notes at the Cloud and Huang related-work bullets (cold-reader
panel): lead Cloud with parameter-vs-activation space; state Huang's
keep-vs-remove inversion plainly; flag the unmeasured hack-basis==clean-basis
question as a reviewer attack vector.
- Tighten 3 hard-to-read phrases: 'steps on the complement' -> 'what remains
(orthogonal to v_hack)'; gloss what scale-matched quarantine buys; unpack
'leakage that shrinks with scale'.
- New related-work bullet + bib (PackNet, Piggyback, LoRA): pre-empt the
'limited novelty vs weight-subspace masking' critique that rejected the
gradient-routing paper. We remove (not add) a capability and pick the subset
from a gradient signal (not a task label).
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
route2 is an internal run-tag, not something a reader cares about.
Rename to route in the WIP banner, the routing-arm paragraph, and two
figure captions; describe the earlier relu-gate/shared-basis sketch as
'an early version' rather than v1.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
User settled it: prog_wide pairs were AI-authored (Claude), so the
synthetic/AI-written framing in contribution 2 is honest. Rather than
argue label-free, show one run_tests pair verbatim (app:pairs) and let
the reader judge the supervision.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Huang related-work bullet now states the actual differences (SVD of clean
update trajectory + warmup vs our contrastive pair-gradients in delta_S coords;
they project onto trusted, we project out hack; we quarantine+delete at deploy,
they only constrain training). Renamed docs/papers/grad_routing/paper_deng_*
-> paper_huang_* (untracked note; correct attribution is Huang et al. 2026).
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
- title -> 'Quarantining Reward-Hacking Gradients with a Hacking Representation'
- contributions: (1) adapt SGTM parameter-gradient masking from supervised
unlearning to RL reward hacking, route+ablate framing from gradient routing
but NOT Cloud's activation .detach(); (2) replace the data-label mask with a
RepE-extracted hack direction from ~10-21 pairs (live rollouts unlabeled).
- method 'Arms': call route SGTM-style post-backward parameter masking in SVD
basis, routed into a deletable subspace.
- related work: Cloud = localize-then-ablate idea only; SGTM = closest
mechanistic relative, their TPR/FPR knob = our weak-detector axis.
- title comment flags the OPEN synthetic-pairs question (headline v_hack is
hand-authored prog_wide, not AI-prompted).
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
- route2 -> route in all prose/captions/tables (route2 stays in % provenance
comments as the run-tag). A reader does not care about the version number.
- title: steering-vector framing; recorded naming reasoning as a comment (do NOT
claim label-free -- our pairs ARE labels; the backable scoped claim is held-out
hacks suppressed with zero labels of their own, earnable by A5).
- FLAG at contribution 1: our mechanism is SGTM-style post-backward parameter-
gradient masking, NOT Cloud's activation-level gradient routing. Author-verbatim
claim left intact but flagged inline; see docs/papers/grad_routing/sgtm_vs_ours.md.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Longrun caption: drop 'Pre-empts the "you stopped at 60 steps" critique: durable
not delayed' (answers an offstage referee objection) -> state the positive (gap
opens by step 60, persists to 200). Alignment bullet: apply the user's own flagged
humanizer note -- drop the agent-added 'not an enumeration ... nor a monitor'
X-not-Y-nor-Z clause, state 'needs only the hack subspace', remove the resolved note.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
make_random_vhack: drop 'run 31's v_hack' dead run-ID + 'the task number', state
the actual role (shape source / reproducible seed). probe_distill: collapse the
'either finish or remove, for now...' train-of-thought into a tight FIXME.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
verify_rewards, verify_vhack_heldout, build_substrate, probe_distill, probe_plot_stack
are run via 'python -m' / justfile and imported by no core module -> moved to scripts/,
relative imports rewritten to 'from projected_grpo.X'. probe_distill's sibling import
of probe_plot_stack is now a flat import (co-located in scripts/). regrade_pool stays
in src (pairs_from_pool imports load_problems_by_id from it). justfile recipes updated.
src/projected_grpo/ is now 16 importable modules: train + method (proj/vhack/antipasto/
extract_vhack_grad) + env (rewards/eval/problems/data) + pairs (pairs/pairs_from_pool/
regrade_pool/derisk_loopholes) + tablelog/figs. ~1480 lines moved out of the package.
Smoke green (verify_rewards 52/52 from scripts/, train pipeline cout->0).
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Dropped dead job-ID narrative (job 60/64) on rollout_ablate_frac, the
'vanilla step 17' dead-run ref in eval.py, the 'old signed sum' dead-code ref in
proj.py, and the conversational 'current experiment line' lead. Removed the dead
probe-traj justfile recipe. Kept all TODO/FIXME and the 'why' memory-tuning
comments. Smoke green (cout->0).
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Deleted (zero importers/refs): scripts/{migrate_out_dirs,audit_log,plot_route_evidence}.py
and src/projected_grpo/{bake_lora,probe_lora_runtime,probe_traj}.py (LoRA-merge path
+ dev trajectory comparator, superseded). Removed the dead probe-traj recipe.
Quarantined to scripts/attic/: make_pairsets.py + make_dataset_pairsets.py (persona-pair
authoring, tasks #123-126 done; live path is pairs.PAIRS / pairs_from_pool).
Comments: dropped dead job-ID narrative (job 60/64) on rollout_ablate_frac, the
'vanilla step 17' dead-run ref in eval.py, the 'old signed sum' dead-code ref in
proj.py, and the conversational 'current experiment line' lead. Kept all TODO/FIXME
and the 'why' memory-tuning comments. Smoke green (cout->0).
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
The 2x2 train row used per-step hack_s (noisy n=28 train batch, knob-on) vs the
deploy row's smooth n=64 eval (knob-off) -- different estimators, confounded.
Now at each eval step route arms ALSO run the SAME n=64 eval with the quarantine
ACTIVE (knob-on = training policy), logged as hk_on/slv_on. vanilla/erase reuse
deploy (no quarantine -> knob-on==knob-off). plot_dynamics prefers hk_on for the
train series so the 2x2 differs ONLY in knob state.
Also: plot parses --teacher-off-step from argv and shades the teacher-ON region
[0,toff] + a dashed cut line in the 2x2. The stashed long-run route2 jobs
(92 KL, 94 teacher-off) inherit the knob-on eval automatically at runtime.
Smoke (route2 hk_on present + logged, both plot parse paths) green.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
paper-longrun, paper-noteacher, paper-teacheroff, paper-harvest -- each pueue-adds
with a why:/resolve: label so every paper job is reproducible from one command.
longrun uses the KL-stabilised optimizer (beta=1e-5, Adam 0.9/0.99).
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
The A4 framing in one figure: vanilla train==deploy (cheat in the weights),
route2 train HACKS while deploy is clean (cheat in the deletable knob). parse_log
now keeps the raw train series (hack_train/solve_train) before the deploy
substitution. New fig: dyn_longrun_200_train_deploy.png.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
rollouts.jsonl now carries problem_id, env_mode, and the exact chat-templated
prompt -- the per-prompt problem is a random draw, so these are required to harvest
same-prompt (hack,clean) pairs from real student rollouts (A5 held-out v_grad; the
teacher pool is a different distribution, not IID with student hacks).
--teacher-off-step=N: seed hacks via teacher pool for N steps then cut to pure
on-policy (G_t=0) -- guarantees all hacks emerge before testing route2 persistence
without the teacher crutch. Smoke (curriculum fires at step 2, metadata present) green.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
train.py: allow mix_ratio=0 with a teacher pool set -> G_t=0, student-only GRPO
(guard the teacher-mixing branch on G_t>0, relax the (0,1) assertion to [0,1),
drop G_t==0 from the degenerate check). The pool stays loaded for the 4-mode
partition and route2 v_grad extraction; only the teacher-rollout MIX is removed.
Smoke (mix=0 + normal mix=0.5 + vanilla) all green.
Also: fill A4 long-run figure (fig:longrun) in main.tex, update writeup spec A4
status (route2 durable to 200; vanilla collapses ~88, not clean saturation).
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
plot_dynamics now dumps the plotted series to <out>.csv next to the png so the
figure is reproducible from a tracked artifact (logs/ and out/runs/ are gitignored).
ylim floor -0.035 + 'hack ≡ 0' annotation so route2's pinned-at-zero hack line is
visible rather than hidden under the axis.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
The worktree-refactor merge (a1b17ab) left the pre-refactor StepLogger
(plus _Col/_format_cell) defined in train.py, shadowing the import from
tablelog.py. Call site uses the new mode_code signature, so the shadow
raised TypeError on every run -> jobs 75/76/77/78/84 all died at startup.
The two copies had diverged: train.py's shadow carried the newer
deploy-for-all-arms layout (task #179) + per-mode int columns, while
tablelog's extracted copy had the older routing-only deploy placement +
frac per-mode. Ported the correct layout into tablelog (the refactor
target) and deleted the shadow. Verified via smoke + smoke-vanilla:
vanilla header now shows hk_dep/slv_dep, routing2 has them once,
per-mode hk_rt renders as int.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
- deploy hack/solve is now the headline metric for all arms, so turn the
mid-train deploy-eval on by default (smoke now covers the deploy path too);
200-step runs pass a sparser cadence explicitly.
- docs/spec/20260602_writeup_spec.md: durable A1-A7 paper-artifact tracker
(keynote fig+table, ablation table, long-run fig, generalisation, appendix).
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
scripts/tt_erase_bench.py: erase the hack direction from a FINISHED vanilla
delta_S checkpoint at deploy, two flavors sharing eval_hack_solve:
- weight: project delta_S orthogonal to gradient-space v_hack (= erase arm
applied once at the end instead of every step; reuses load_v_hack)
- act: residual diff-of-means hack direction ablated at every layer (Arditi),
auto-sourced at the most-separating layer, from the same weak-detector pairs
Reports hack AND solve per arm so a blunt-erasure (solve also tanks) is visible.
Baseline for whether train-time routing beats cheap post-hoc erasure.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
- blog: mark as erase-n=2 draft, note route2/exploration-floor/deploy-eval are the
current direction; embed dyn_sub4_hack_overlay.png (force-added); ASCII em-dashes;
de-bold the arm list (#15 tell)
- README: add route2 arm + apples-to-apples deploy-eval to 'What we compare'; stale
banner on the n=1 mix=0.5 findings
- plot_dynamics: remove _mark_if_sparse (asymmetric sparse-only dots); EMA-held line
for all arms
- train.py: fix 'held-out greedy' -> 'held-out eval subset, T=0.7' (deploy eval is
sampled, not greedy)
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Wassname flagged the dynamics curve wasn't comparable: route2 plotted its
deploy eval (n=64, T=0.7, every 5 steps) while vanilla/erase plotted training
rollouts (n=28, every step) -- route2 looked artificially smoother. (NOT a
temperature gap: both gens are T=0.7; the "held-out greedy" header was a stale
lie, now corrected.)
train.py: ungate the periodic DEPLOY-eval to run for EVERY arm. route/route2
wrap it in ablate_quarantine (deploy = knob zeroed); vanilla/erase use
nullcontext (deploy == trained model). Same estimator across arms. Cost: ~+40%
amortized generation on the arms that newly get it (n=64 every 5 steps over
~32 train gens/step) -- n stays 64 to match the finished route2 n=3.
plot_dynamics.py: plot hk_dep/slv_dep for ALL arms when present (drop the
route-only guard; old logs fall back to training hack_s). Drop the cos row
(it was for online-vs-offline erasure; not informative next to the rate row,
and the cross-arm cos comparison was apples-to-oranges) -> single-row small
multiples, "deployed rate". Title states deploy-eval n=64 T=0.7.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
route2's plotted hack/solve is the DEPLOY eval (hk_dep/slv_dep): greedy, n=64,
logged every eval_ablate_every=5 steps and EMA-held flat between. The held line
reads as per-step-dense and oversells route2's smoothness vs the per-step
temperature-sampled (n=28) training curves the other arms plot -- an apples-to-
oranges smoothness the reader shouldn't be misled by.
_mark_if_sparse dots the real measured points when a series is >50% NaN; dense
series (training hack_s, cos sep/leak) stay unmarked. Now the route2 curve
visibly rests on ~13 eval points, not 60.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
The cos row mixed non-comparable quantities: erase logged cin_t/cin_s
(pre-intervention single cosines), route2 logged hkgap (a difference) and
resid (post-intervention), all under one "cos(grad,v_hack)" ylabel. Wassname
flagged it -- they are not the same measurement.
Derive two quantities that mean the same thing in every column (_add_cos_derived):
sep = does v_hack still discriminate hacky from clean gradient
erase: cin_t-cin_s (teacher vs student); route2: hkgap (hackflag vs clean)
leak = residual hack-alignment of the post-intervention deployed gradient
erase: cout (after projection); route2: resid (after routing) -- same quantity
Legend now lands on the leftmost arm that has cos data (vanilla has none).
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Stacked hack (top) + solve (bottom) sharing x; EMA-5; onset dot on hack only;
arms direct-labelled once on solve with y de-collision + leader lines (the three
non-route arms overlap, so their labels would otherwise stack). routing2 reads
hack~0 / solve highest at a glance.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
- plot_dynamics: routing (route v1) out of ARM_ORDER -- superseded by routing2.
- plot_substrate: per-mode hk_* are now plain per-batch counts (streaming log
dropped the /denominator); parse the count, plot it (EMA or cumsum); skip old
n/d-format logs (incompatible units). Y-axis hacks/batch, count annotations.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
wassname