-
b497db97ba
fix: guard floor phase against n_trainval budget + assert epsilon on output
iomgaa
2026-07-15 12:38:33 -04:00
-
c3187167c8
feat: add greedy joint-constrained split selector
iomgaa
2026-07-15 12:32:24 -04:00
-
84b0b30213
fix: dedup three-bucket projection + honest C3 error-handling note
iomgaa
2026-07-15 12:29:58 -04:00
-
aa10485b9f
fix: make pools.json freeze atomic + add split manifest
iomgaa
2026-07-15 12:28:48 -04:00
-
fd907aab46
fix: extend correctness fail-fast to test-side pool questions (P5)
iomgaa
2026-07-15 12:22:11 -04:00
-
53989078a0
feat: add offline baseline diagnosis orchestration
iomgaa
2026-07-15 12:22:07 -04:00
-
3d8bd75372
feat: build all-video records with difficulty and signal overlay
iomgaa
2026-07-15 12:18:56 -04:00
-
20eea98cdd
refactor: add video-atomic pool split (algo #5 gate input preserved)
iomgaa
2026-07-15 12:14:36 -04:00
-
9d19328cc9
feat: add signal tiering
iomgaa
2026-07-15 12:12:48 -04:00
-
75a2aac1c0
test: cover unknown error_type ValueError path
iomgaa
2026-07-15 12:10:24 -04:00
-
0a7ba724a9
feat: add evolution_target derivation and cell
iomgaa
2026-07-15 12:06:56 -04:00
-
3a8dd5c167
fix: add close/context-manager to diagnosis store; tighten import boundary
iomgaa
2026-07-15 12:06:29 -04:00
-
6b85fcbe1c
feat: add baseline diagnosis signal store
iomgaa
2026-07-15 12:01:30 -04:00
-
01f2e7c7b9
feat: add steps_json-backed RunLog wrapper
iomgaa
2026-07-15 11:51:55 -04:00
-
cf7f15d8bb
feat: add steps_json to trace-row adapter
iomgaa
2026-07-15 11:47:00 -04:00
-
745b600b5d
docs: register baseline_diagnosis and split_manifest schemas with baseline metrics
iomgaa
2026-07-15 11:42:38 -04:00
-
cfa510a7e5
docs: add results-driven video-split implementation plan
iomgaa
2026-07-15 11:38:32 -04:00
-
312cee7be7
docs: add results-driven video-level split design
iomgaa
2026-07-15 11:17:45 -04:00
-
f8ad8f0143
test: harden baseline_cache guard to per-call unit_id check
iomgaa
2026-07-15 10:13:33 -04:00
-
a536a81b0d
chore: apply ruff format to dead-path guard test
iomgaa
2026-07-15 10:07:19 -04:00
-
13234426b5
chore: guard perquestion revival + mark v2 tables deprecated
iomgaa
2026-07-15 10:06:13 -04:00
-
e9349b55c0
test: add v3 pair-contract full-chain integration regression
iomgaa
2026-07-15 09:57:39 -04:00
-
58a84dc72b
refactor: add CHECK constraints and drop redundant index in v3 tables
iomgaa
2026-07-15 09:47:00 -04:00
-
3cc8dc9105
refactor: assert warn sink, rename disk-pair sifter, trim WHAT docstrings
iomgaa
2026-07-15 09:42:27 -04:00
-
5ecbac620c
feat: add v3 question-gen observation tables to run_store
iomgaa
2026-07-15 09:33:40 -04:00
-
1ecb6ba25c
fix: fail-loud on disk-corrupt pairs and preserve buffer on reject
iomgaa
2026-07-15 09:31:42 -04:00
-
bdcc93d7de
feat: add pair atomic accepted-bank writer helper
iomgaa
2026-07-15 09:22:41 -04:00
-
19911e18e0
test(harness): 黄金测试加非空护栏、补类型注解、端到端对照旧逻辑
iomgaa
2026-07-15 08:28:12 -04:00
-
6c6fb576ee
feat(harness): checkpoint 存 unit_id 序列，断点续跑孪生对不拆
iomgaa
2026-07-15 08:22:18 -04:00
-
bd1f7a22a2
feat(harness): pools.json 序列化 pair 四字段防孤儿 single
iomgaa
2026-07-15 08:05:15 -04:00
-
273984674b
feat(harness): 迁移 gate 信息量阶梯到 unit 粒度 + schema_version
iomgaa
2026-07-15 07:56:15 -04:00
-
7e97081779
test(harness): 补 _ladder_units 直测 + 澄清 gate 观测表 unit_id 口径注释
iomgaa
2026-07-15 07:40:36 -04:00
-
4b6d1d8a50
feat(harness): correctness 三口径 + gate 块按 unit 跑
iomgaa
2026-07-15 07:31:03 -04:00
-
dee6bf4896
test: assert specific missing question_id in T6 invariant error match
iomgaa
2026-07-15 07:09:54 -04:00
-
8a69a54078
refactor: fail-loud unit grading and stricter orphan pair drop
iomgaa
2026-07-15 07:05:37 -04:00
-
61bfa0e633
style: apply ruff format to T6 inference aggregation
iomgaa
2026-07-15 06:58:51 -04:00
-
730caa7e9a
feat: aggregate inference by question unit with pair AND
iomgaa
2026-07-15 06:57:48 -04:00
-
2429dad393
feat(batching): unit 粒度切分——pair 整锁 + 单元级分桶 + 非 AR 独立 rng
iomgaa
2026-07-15 06:46:28 -04:00
-
c412698cff
style(test): 修 loader unit-sampling 测试的 ruff 告警
iomgaa
2026-07-15 06:30:14 -04:00
-
d6a3107e4e
feat(question_gen): loader 按 unit 分层采样 + load_benchmark 读回 pair 字段
iomgaa
2026-07-15 06:27:41 -04:00
-
ddb9a44f75
feat(pools): 三池切分以 unit 为原子，孪生对同池不被拆散
iomgaa
2026-07-15 06:09:14 -04:00
-
5ef5f2b8b7
refactor: extract _assemble_pair to lower build_units complexity to B
iomgaa
2026-07-15 05:54:08 -04:00
-
bef46636fe
feat: add question_units helper as pair contract entry point
iomgaa
2026-07-15 05:52:30 -04:00
-
7ef9b99217
feat: add QuestionUnit entity and GeneratedQuestion pair fields
iomgaa
2026-07-15 05:45:34 -04:00
-
0fe1c96393
docs: add question-gen v3 construction-paradigm design and phase1 plan
iomgaa
2026-07-15 05:41:10 -04:00
-
-
d9f7dee2df
docs: revise question-gen v2 design per Codex review (pair-unit contract, AR-only gate, phasing)
iomgaa
2026-07-15 01:21:28 -04:00
-
7949f16906
docs: add question-gen v2 grounded-contrastive design
iomgaa
2026-07-15 01:10:00 -04:00
-
fd96f906c2
docs: add question-gen v2 diagnosis and locked redesign direction
iomgaa
2026-07-15 00:50:29 -04:00
-
49b7be78ae
fix: derive OUTPUT from config output_dir so CONFIG override counts correctly
iomgaa
2026-07-14 21:12:48 -04:00
-
d84f9bf777
chore: harden generate_ar30.sh (drop no-op mock, add zero-gain early stop)
iomgaa
2026-07-14 21:08:00 -04:00
-
b1c1bf7aac
fix: isolate selector VLM exceptions at slot level (no batch crash)
iomgaa
2026-07-14 21:08:00 -04:00
-
441a0aa6c3
chore: format Phase B modules and register Phase B plan in wiki
iomgaa
2026-07-14 17:11:04 -04:00
-
f36eb66c18
fix: tolerate backfill under-delivery, fix predict None-fill, drop dead session_id
iomgaa
2026-07-14 17:07:55 -04:00
-
1d222d9f18
feat: wire real agent runner, backfill assembly and adversarial-filter CLI
iomgaa
2026-07-14 16:58:32 -04:00
-
73d0bb9190
feat: add adversarial round loop with backfill iteration and difficulty report
iomgaa
2026-07-14 16:40:05 -04:00
-
c1565a01c2
test: assert stage/verdict/pair_id and Q-refusal flip_skipped path
iomgaa
2026-07-14 16:33:09 -04:00
-
15aee0cfc1
feat: add flip gate reusing P prediction and mirror-Q agent run
iomgaa
2026-07-14 16:26:51 -04:00
-
8731e448fe
fix: harden mirror material rebuild (empty/invalid frames, dedup, observability)
iomgaa
2026-07-14 16:16:38 -04:00
-
4af99b700b
feat: add mirror question generation with canonical distinctness check
iomgaa
2026-07-14 16:08:13 -04:00
-
d4e9852864
feat: add cheater gate with resume-safe survivor recovery
iomgaa
2026-07-14 16:00:14 -04:00
-
24ed7ca322
fix: harden canonical_answer_text and align decision-core signatures to spec
iomgaa
2026-07-14 15:55:36 -04:00
-
c109f2257a
feat: add pure decision core for adversarial filter (hash/fingerprint/canonical/flip)
iomgaa
2026-07-14 15:51:28 -04:00
-
334fbbc94d
feat: add AdversarialFilterConfig for Phase B post-hoc filter layer
iomgaa
2026-07-14 15:47:21 -04:00
-
ac115d96fb
test: cover backfill params end-to-end (mutation isolation + seq_offset)
iomgaa
2026-07-14 15:44:43 -04:00
-
8b9e8aa19f
feat: add optional backfill params to run_pipeline_v2
iomgaa
2026-07-14 15:39:57 -04:00
-
d77cbc95eb
feat: add adversarial_verdicts table with resume and terminal-verdict query
iomgaa
2026-07-14 15:35:11 -04:00
-
e41a2b0d08
test: guard non-AR flip isolation and AR flip-capable set
iomgaa
2026-07-14 15:30:18 -04:00
-
f12dd7e559
feat: add supports_flip/flip_axis to SubPattern for Phase B flip gate
iomgaa
2026-07-14 15:27:21 -04:00
-
841112c6af
docs: revise Phase B plan per Codex review (C1-C3,I1-I6,M1-M3)
iomgaa
2026-07-14 15:24:44 -04:00
-
abc9c097d2
docs: add Phase B adversarial question-gen implementation plan
iomgaa
2026-07-14 15:05:36 -04:00
-
96e314c3a0
fix: mark item rejected on selector_error for consistent bookkeeping
iomgaa
2026-07-14 14:49:41 -04:00
-
46ac848176
fix: converge non-numeric selector scores to ValueError
iomgaa
2026-07-14 14:47:21 -04:00
-
eaab865d84
docs: register Phase A plan in research wiki
iomgaa
2026-07-14 14:41:53 -04:00
-
e64c26e578
style: format distractor_selector
iomgaa
2026-07-14 14:41:21 -04:00
-
76f719018c
feat: loosen multi_true gate to qualifier-scoped correctness
iomgaa
2026-07-14 14:33:22 -04:00
-
b1f15ddb3a
fix: keep cross_segment rule single-dimension, guard no-absent-events
iomgaa
2026-07-14 14:31:30 -04:00
-
58278c6de4
feat: enforce single-dimension counterfactual in AR distractor rules
iomgaa
2026-07-14 14:28:32 -04:00
-
ea38e4717b
refactor: extract grounded selector application into helper
iomgaa
2026-07-14 14:22:30 -04:00
-
b13eab0659
feat: wire grounded selector into AR slot processing
iomgaa
2026-07-14 14:15:50 -04:00
-
d0194f5840
fix: degrade distractor pool gracefully on malformed VLM response
iomgaa
2026-07-14 14:11:13 -04:00
-
8a54055d02
feat: add grounded distractor selector with visual scoring
iomgaa
2026-07-14 14:06:07 -04:00
-
207e834f30
feat: add selector_scores observation column to question_gen_items
iomgaa
2026-07-14 13:59:29 -04:00
-
3f984acc18
feat: add uses_grounded_selector strategy switch (AR only)
iomgaa
2026-07-14 13:53:17 -04:00
-
2608a3841f
style: reformat sub_pattern round-trip tests
iomgaa
2026-07-14 13:51:51 -04:00
-
e68e4b7d57
fix: restore sub_pattern when loading benchmark JSON
iomgaa
2026-07-14 13:51:05 -04:00
-
111c88488f
style: format sub_pattern test file
iomgaa
2026-07-14 13:46:25 -04:00
-
ae0a718f67
feat: thread and persist sub_pattern into accepted questions
iomgaa
2026-07-14 13:45:31 -04:00
-
25f2a845ff
fix: run_gates must use current_tree after video resample
iomgaa
2026-07-14 13:40:46 -04:00
-
92714db370
docs: revise Phase A plan per Codex review (C1-C2,I1-I5,M1)
iomgaa
2026-07-14 13:38:35 -04:00
-
645585e784
docs: add Phase A grounded question-gen implementation plan
iomgaa
2026-07-14 13:28:43 -04:00
-
ad0bfe3a51
docs: add grounded + adversarial question-gen design (Phase A/B)
iomgaa
2026-07-14 13:21:06 -04:00
-
84b52a0311
feat(pools): auto-supplement maintenance correct questions in PerCategoryPoolStrategy
iomgaa
2026-07-14 10:41:46 -04:00
-
72befa2bd4
feat(pools): add batch_correct_ratio field to PoolConfig
iomgaa
2026-07-14 10:34:28 -04:00
-
453cf62088
docs: add maintenance pool implementation plan
iomgaa
2026-07-14 10:22:52 -04:00
-
73c5d1e380
docs: add high-correctness warning to maintenance pool design
iomgaa
2026-07-14 10:13:57 -04:00
-
89cad891de
docs: add maintenance pool auto-supplement design
iomgaa
2026-07-14 10:11:26 -04:00
-
d7d7ce5bdc
feat(question_gen): register ActionRecognitionStrategy, replace temp VISUAL binding
iomgaa
2026-07-14 06:51:50 -04:00
-
d0a8019fe1
feat(question_gen): add ActionRecognitionStrategy with 6 SubPatterns
iomgaa
2026-07-14 06:49:03 -04:00
-
eaa58604b3
feat(question_gen): add Action Recognition specialized prompt template
iomgaa
2026-07-14 06:45:09 -04:00
-
b976c789d2
docs: add Plan B — ActionRecognitionStrategy implementation plan
iomgaa
2026-07-14 06:34:40 -04:00