wassname fb9f68530c refactor: move pair data out of pairs.py into build script; drop tainted axis-1/3 ...

pairs.py now only has HackPair dataclass + _prompt/_wrap helpers.
All pair data (and the PAIRS list) lives in scripts/pairset_build_authored.py.

Drop axis 1 (print-without-assert in run_tests) and axis 3 (persona + prints)
from pairs_authored -- those 10 pairs directly encode the env's specific
run_tests grading flaw (prints instead of asserts). Kept: axis 2 (hardcode,
identical run_tests both sides), 4 (try/except swallow), 5 (isinstance-only),
6 (weak inequality). P19-P21 (previously defined but unused) now included.
Result: 11 pairs from 4 axes.

Also removed the PAIRS-fallback branch from extract_vhack_grad.py (same
fail-loud pattern applied to train.py earlier).

Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>

2026-06-09 08:44:33 +00:00

..

attic

retract 'null_city contaminated' framing -> in/out-of-subspace + cosine-is-correlational

2026-06-05 09:21:41 +00:00

build_combined_pool.py

reorg: out/ sorted by datatype (vhack/ pools/ runs/ vhack_grads/ figs/)

2026-05-30 03:52:24 +00:00

build_runtests_pool.py

fix: dense run_tests teacher pool (6 -> 215 prompts) so the hack seeds in 60 steps

2026-06-07 11:01:31 +00:00

build_substrate.py

rename python package projected_grpo -> vgrout

2026-06-05 14:51:48 +08:00

diag_cosine_dist.py

pairs: de-confound v2 (print(==) vs assert, line-matched) + intent designs (think/funcname/concept)

2026-06-08 13:08:47 +00:00

diag_pairs_compare.py

chore: memory updates, diag_pairs_compare script

2026-06-09 02:42:56 +00:00

make_random_vhack.py

cleanup: trim 2 stale provenance/train-of-thought comments

2026-06-03 00:25:22 +00:00

pairs_from_rollouts.py

rename python package projected_grpo -> vgrout

2026-06-05 14:51:48 +08:00

pairset_build_authored.py

refactor: move pair data out of pairs.py into build script; drop tainted axis-1/3

2026-06-09 08:44:33 +00:00

pairset_build_progsets.py

refactor: named pairset JSONs + explicit --vhack-pairs-path, remove None fallback

2026-06-09 08:09:09 +00:00

plot_deploy_overlay.py

rename python package projected_grpo -> vgrout

2026-06-05 14:51:48 +08:00

plot_dynamics.py

rename python package projected_grpo -> vgrout

2026-06-05 14:51:48 +08:00

plot_emergence.py

rename python package projected_grpo -> vgrout

2026-06-05 14:51:48 +08:00

plot_substrate.py

rename python package projected_grpo -> vgrout

2026-06-05 14:51:48 +08:00

probe_distill.py

rename python package projected_grpo -> vgrout

2026-06-05 14:51:48 +08:00

probe_plot_stack.py

refactor: move 5 leaf entrypoints src/ -> scripts/ (src is now library-only)

2026-06-03 00:23:56 +00:00

rescore_deploy.py

fix: eval on paper test set, not contaminated holdout (base solve 0.94->0.094)

2026-06-07 11:01:31 +00:00

results_deploy.py

feat: online_stats gate for routeV -- live q5/q95 band calibration

2026-06-09 02:25:37 +00:00

results.py

results: just results = eval2 deploy table (time/headline/deploy/arm/pair/seed/train/argv); hard eval2 cutoff; archive eval1 (Q1-Q13 + 352 old logs)

2026-06-09 01:50:42 +00:00

tt_erase_bench.py

rename python package projected_grpo -> vgrout

2026-06-05 14:51:48 +08:00

validate_spoonfeed.py

rename python package projected_grpo -> vgrout

2026-06-05 14:51:48 +08:00

verify_base_solve.py

fix: eval on paper test set, not contaminated holdout (base solve 0.94->0.094)

2026-06-07 11:01:31 +00:00

verify_eval_gap.py

eval: train/test token gap for all 4 modes (lenient disjoint families)

2026-06-06 13:49:07 +00:00

verify_partition.py

test: no-cheat partition + teacher-pool composition gate (verify_partition.py)

2026-06-05 04:36:03 +00:00

verify_rewards.py

rename python package projected_grpo -> vgrout

2026-06-05 14:51:48 +08:00

verify_vhack_heldout.py

rename python package projected_grpo -> vgrout

2026-06-05 14:51:48 +08:00