Kral 229f862600 Generator training mode: train pool, eval overlap check, category mix, error-targeted tasks ...

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

2026-10-05 11:53:07 +02:00

..

__init__.py

Harness: runner, generator, tasks T01 T13 T14 T15, CLAUDE.md

2026-10-02 21:42:51 +02:00

adt_client.py

ADT fallback: Accept headers for FUNC; G0189 accepted; regen2 results

2026-10-03 16:31:03 +02:00

agents.py

Loop guard: also identical call with identical result 3 times in a row; baseline restarts (run base 20500)

2026-10-04 07:59:28 +02:00

cli.py

Harness: runner, generator, tasks T01 T13 T14 T15, CLAUDE.md

2026-10-02 21:42:51 +02:00

empirical.py

Empirical filter runner (DeepSeek); real cost ratio in docs

2026-10-03 09:29:20 +02:00

evalset.py

In-place E/I: G2 needs a changed seed contract object, G4 skips it; regenerate G0009/G0013/G0016/G0023 as G0188-G0191

2026-10-03 06:33:33 +02:00

generator.py

Generator training mode: train pool, eval overlap check, category mix, error-targeted tasks

2026-10-05 11:53:07 +02:00

judge.py

H and K tested (G0168, G0178 accepted); keyword fast path needs all groups

2026-10-03 06:04:36 +02:00

ledger.py

Harness: runner, generator, tasks T01 T13 T14 T15, CLAUDE.md

2026-10-02 21:42:51 +02:00

mcp_client.py

Harness: runner, generator, tasks T01 T13 T14 T15, CLAUDE.md

2026-10-02 21:42:51 +02:00

mutation.py

Mutation check done on all accepted tasks; mutant fixes; dump filter in proxy

2026-10-03 06:02:59 +02:00

overlap.py

Generator training mode: train pool, eval overlap check, category mix, error-targeted tasks

2026-10-05 11:53:07 +02:00

pilot.py

Mutation in generator; category H (stop) scoring and judge; category K variants and tool schema variant; evalset plan

2026-10-03 05:37:20 +02:00

proxy.py

Devstral Small 2: serve.sh, baseline.py without thinking args, repair rate, activation error message fix, stop rule in chain (run base 21000)

2026-10-04 08:57:12 +02:00

review.py

Eval review v1: checklist, review helper, 24 tasks reviewed; G0002/G0022 tests fixed; I/E generator definitions

2026-10-03 06:23:38 +02:00

runner.py

baseline: setup failure is not recorded as a result; repair_stats tolerates a missing trajectory

2026-10-04 12:07:33 +02:00

task.py

Harness: runner, generator, tasks T01 T13 T14 T15, CLAUDE.md

2026-10-02 21:42:51 +02:00

trainset.py

Generator training mode: train pool, eval overlap check, category mix, error-targeted tasks

2026-10-05 11:53:07 +02:00