config: eval2/eval3 regime + fold per-token into arm + FastLora; drop contaminated prog_wide; OOD pairs

Browse Source

Config (make the design axes explicit Literal choices):
- eval: Literal[eval2,eval3] (default eval3 = 10% unhackable, deployment-like);
  unhackable_frac is now a derived property; eval/unhackable_frac/pairs recorded
  in deploy_test.json metadata.
- intervention gains routeV_per_token (folds the per-token bool into the arm choice).
- routeV_gate documented as the pinning axis.
- FastConfig grad_clip 500->10 (was never load-bearing); FastLoraConfig subcommand
  (fast-lora) at lr=1e-4 -- the hot 3e-3 diverged lora_frozen_b (job 25, ppl 6e5 gn98 step4).

Pairs:
- delete prog_wide.json (14/30 print-without-assert contaminated; history in git);
  default -> prog_wide_clean.
- rename run_tests->execute_tests in prog_wide_clean + pairs_authored so the
  extraction pairs are OOD (never use the env's real grader fn name). Re-extracted
  v_hack_smoke to match.

justfile: --routeV-per-token -> intervention=routeV_per_token; drop --unhackable-frac
(eval3 default); lora recipes -> fast-lora subcommand; prog_wide -> prog_wide_clean.

smoke green (erase + routeV_per_token); all 4 verify gates pass.

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

...

This commit is contained in:

wassname

2026-06-10 04:21:54 +00:00

co-authored by Claudypoo

parent b36e3db255

commit 7f45189f1c

7 changed files with 53 additions and 30 deletions

Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL

Download Patch File Download Diff File Expand all files Collapse all files

out/pairsets/pairs_authored.json

+1 -1

File diff suppressed because one or more lines are too long

out/pairsets/prog_wide.json

-1

File diff suppressed because one or more lines are too long

out/pairsets/prog_wide_clean.json

+1 -1

File diff suppressed because one or more lines are too long

out/vhack/v_hack_smoke.safetensors

BIN

Binary file not shown.