wassname caa0d09472 broad: TEACHER_RT -> dense pool (was sparse, under-seeds); log: rename table cols train/deploy (drop 'knob') ...

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

2026-06-07 22:12:00 +00:00

..

__init__.py

rename python package projected_grpo -> vgrout

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

antipasto.py

refactor: drop shadowed-import + duplicate-definition cruft (-91 LOC)

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

data.py

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

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

derisk_loopholes.py

rename python package projected_grpo -> vgrout

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

eval.py

eval+env: single-mode run_tests, held-out val/test eval, both hack metrics

2026-06-07 03:07:35 +00:00

extract_vhack_grad.py

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

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

figs.py

rename route2->routeV; heavy seeded final eval; save delta_S_hack

2026-06-06 12:08:28 +00:00

pairs_from_pool.py

rename python package projected_grpo -> vgrout

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

pairs.py

rename python package projected_grpo -> vgrout

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

problems.py

rename python package projected_grpo -> vgrout

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

proj.py

rename python package projected_grpo -> vgrout

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

regrade_pool.py

rename python package projected_grpo -> vgrout

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

rewards.py

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

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

tablelog.py

log: routeV routing as keep/resid/rout zones x unit+energy views; drop dead hk_abl/slv_abl

2026-06-07 13:13:01 +00:00

train.py

broad: TEACHER_RT -> dense pool (was sparse, under-seeds); log: rename table cols train/deploy (drop 'knob')

2026-06-07 22:12:00 +00:00

vhack.py

rename python package projected_grpo -> vgrout

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