Hugging Face's logo

Datasets:
ErenJaegerYeager
/
ClawBenchPro

Tasks:
Text Generation
Question Answering
Modalities:
Text
Formats:
csv
Languages:
Chinese
English
Size:
1K - 10K
Tags:
agent-benchmark
workplace
tool-use
multi-turn
skills
nanoclaw
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
 Community
1
ErenJaegerYeager commited on
Commit
dee1239
·
verified ·
1 Parent(s): d3b76fa

Add workplace verifiers for ClawBenchPro

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. checksums.sha256 +0 -0
  2. manifest.json +4 -4
  3. persona_aligned_mix_200/checksums.sha256 +105 -104
  4. persona_aligned_mix_200/manifest.json +2 -2
  5. persona_aligned_mix_200/provenance/verifier_materialization_manifest.jsonl +3 -0
  6. persona_aligned_mix_200/provenance/verifier_repair_manifest.jsonl +2 -2
  7. persona_aligned_mix_200/tasks/data_persona_aligned_base_50_0027/verify_workplace.py +27 -144
  8. persona_aligned_mix_200/tasks/data_persona_aligned_base_50_0028/verify_workplace.py +26 -259
  9. persona_aligned_mix_200/tasks/data_persona_aligned_base_50_0031/verify_workplace.py +26 -121
  10. persona_aligned_mix_200/tasks/data_persona_aligned_base_50_0038/verify_workplace.py +27 -192
  11. persona_aligned_mix_200/tasks/data_persona_aligned_base_50_0039/verify_workplace.py +26 -142
  12. persona_aligned_mix_200/tasks/data_persona_aligned_base_50_0044/verify_workplace.py +4 -4
  13. persona_aligned_mix_200/tasks/data_persona_aligned_base_50_0050/verify_workplace.py +4 -4
  14. persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0003/verify_workplace.py +26 -96
  15. persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0005/verify_workplace.py +26 -87
  16. persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0008/verify_workplace.py +54 -51
  17. persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0010/verify_workplace.py +27 -109
  18. persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0018/verify_workplace.py +26 -133
  19. persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0025/verify_workplace.py +26 -120
  20. persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0026/verify_workplace.py +27 -162
  21. persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0029/verify_workplace.py +24 -161
  22. persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0031/verify_workplace.py +26 -121
  23. persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0039/verify_workplace.py +26 -142
  24. persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0041/verify_workplace.py +27 -92
  25. persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0042/verify_workplace.py +26 -126
  26. persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0044/verify_workplace.py +4 -4
  27. persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0045/verify_workplace.py +27 -172
  28. persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0049/verify_workplace.py +28 -60
  29. persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0050/verify_workplace.py +4 -4
  30. persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0001/verify_workplace.py +28 -61
  31. persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0002/verify_workplace.py +28 -71
  32. persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0003/verify_workplace.py +26 -96
  33. persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0004/verify_workplace.py +28 -72
  34. persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0005/verify_workplace.py +26 -87
  35. persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0006/verify_workplace.py +26 -111
  36. persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0007/verify_workplace.py +27 -108
  37. persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0008/verify_workplace.py +28 -54
  38. persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0009/verify_workplace.py +26 -207
  39. persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0010/verify_workplace.py +27 -109
  40. persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0011/verify_workplace.py +28 -67
  41. persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0012/verify_workplace.py +27 -103
  42. persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0013/verify_workplace.py +26 -101
  43. persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0014/verify_workplace.py +28 -86
  44. persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0015/verify_workplace.py +27 -106
  45. persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0016/verify_workplace.py +27 -152
  46. persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0017/verify_workplace.py +26 -115
  47. persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0018/verify_workplace.py +26 -133
  48. persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0019/verify_workplace.py +27 -99
  49. persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0020/verify_workplace.py +27 -148
  50. persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0021/verify_workplace.py +27 -206
checksums.sha256 CHANGED
The diff for this file is too large to render. See raw diff
 
manifest.json CHANGED
@@ -16,8 +16,8 @@
16
  "multi_turn_aligned": 200,
17
  "skills_aligned": 200
18
  },
19
- "files": 6359,
20
- "bytes": 23850515,
21
  "checksums": "round_01_aligned_mix_800/checksums.sha256",
22
  "verifiers": "round_01_aligned_mix_800/verifiers"
23
  },
@@ -32,8 +32,8 @@
32
  "multi_turn": 50,
33
  "skills": 50
34
  },
35
- "files": 1397,
36
- "bytes": 6118979,
37
  "checksums": "persona_aligned_mix_200/checksums.sha256",
38
  "verifiers": "persona_aligned_mix_200/verifiers"
39
  }
 
16
  "multi_turn_aligned": 200,
17
  "skills_aligned": 200
18
  },
19
+ "files": 6360,
20
+ "bytes": 24583501,
21
  "checksums": "round_01_aligned_mix_800/checksums.sha256",
22
  "verifiers": "round_01_aligned_mix_800/verifiers"
23
  },
 
32
  "multi_turn": 50,
33
  "skills": 50
34
  },
35
+ "files": 1398,
36
+ "bytes": 6266949,
37
  "checksums": "persona_aligned_mix_200/checksums.sha256",
38
  "verifiers": "persona_aligned_mix_200/verifiers"
39
  }
persona_aligned_mix_200/checksums.sha256 CHANGED
@@ -10,7 +10,7 @@ ddc9ea1b1b06f971daa332d2864fa4a1643dc377693181f75897d2c56056e9af eval_manifests
10
  937eaca759b3664669c8aeb3a0705f12d2fa66a1cb40674e425dbb97048064f4 eval_manifests/skills.jsonl
11
  2d0a464c28dc1aa380cf5740a7ba9aa76bcc33649e46ea32d8ce12ba6b601027 eval_manifests/skills.task_ids
12
  273c0a3d6342edf14abd7ea4f10f9c9179e7982455e8d99be39c0739689d50fa import_manifest.jsonl
13
- f1324a4ff79b58dc8e3e0d9dc8012627fb820e598284db40022c219dec8d5adf manifest.json
14
  12ae83d267551b1e73808354b802a7d0efcc1a3b76453e7b84c9964c4e294503 provenance/eval_manifests/base.jsonl
15
  72aeea0c6321b55982263dbd1cbc23ff114768b3cc21b3cfeab7ff70b7e00284 provenance/eval_manifests/base.task_ids
16
  cdfe914540244feb618a00470b455aba9622d94761352a174dda05826f79d040 provenance/eval_manifests/hard.jsonl
@@ -28,7 +28,8 @@ f886fe8dcee33c3f7ed31e47edea105b4ac16383a7eca79fe2a1b3f584deaa15 provenance/imp
28
  cf3ca3b84a84ca57b8914d4fc3d08e15d04838687ae503baef3206c00888d9ac provenance/selection_summary.json
29
  48fdd4735d4a5bae570f6436e1cfcfe10ba5d236c6522da69ec2960b115670a2 provenance/task_manifest.csv
30
  46c5f60e5218f57b7dd190fee99eb81ca932e52f3f09b11fbf1b76861fa2ef9a provenance/validation_report.json
31
- 77c36f2054abdaf4ff577ee9dc1e04ab6352ce1761c1720aab69e95125b4a29d provenance/verifier_repair_manifest.jsonl
 
32
  7b9957d6b41f006baa0fb5661a7b84520eaeaf2ca0baba12968c99e6e7039033 selection_manifest.jsonl
33
  6b74f75513fad3b4fd1cdebd1e2edc931602987fb6305045f614087f917354c3 skills/data-persona-aligned-skills-50-0001-legacy-raft-parser-skill/SKILL.md
34
  96ea388e187a635832d43c3306cb4c9988d57aed2cf144a92244875fb8c98567 skills/data-persona-aligned-skills-50-0001-legacy-raft-parser-skill/legacy_raft_parser_skill.py
@@ -396,11 +397,11 @@ f5239e4cce892bd9423347e512e9f6bf27a56b4269507834c9efae8c06881b03 tasks/data_per
396
  419c473e36b05d09253cc080d90514eb79e1730c08b20d5c25a28926f4e0e976 tasks/data_persona_aligned_base_50_0026.yaml
397
  309fe08d3f1f04be1366ea65740e959b651f91fe55d4bd13da4c41db1d679414 tasks/data_persona_aligned_base_50_0027/_env_builder_impl.py
398
  701e904057b4ff9c724910458106eee93f4f7fc03fe7469aeee72abdecaccdfd tasks/data_persona_aligned_base_50_0027/env_builder.py
399
- d182de48eacc413d96966204d22d3f85d5dbdf6d09d717ceb32db50076b4e2b1 tasks/data_persona_aligned_base_50_0027/verify_workplace.py
400
  adf7bda9cf224903ae10b3946e8b5ceddd73f5e6d8647f3b12f0fb1d1b5efd94 tasks/data_persona_aligned_base_50_0027.yaml
401
  99e50d5dac2bf083625b694a59f8a0f6788fef363189e9f628af378e2db19e49 tasks/data_persona_aligned_base_50_0028/_env_builder_impl.py
402
  30376ea0cf4aee7c2dec63117aa45844e1fa81511e864a4f65b0dee849272660 tasks/data_persona_aligned_base_50_0028/env_builder.py
403
- 1ff86299dbe258b4d5d5a6ef331188d97aeb437ec121db349e5650fbb36ce067 tasks/data_persona_aligned_base_50_0028/verify_workplace.py
404
  f836f191afc149f21bd1a50c7babd2eaee3c191ecd24415184456dd0209a6115 tasks/data_persona_aligned_base_50_0028.yaml
405
  c1b63bdd52c37439a150793741e470c9ef81c1963175ce5c3b76698704154e21 tasks/data_persona_aligned_base_50_0029/_env_builder_impl.py
406
  1cae3b31de8df3ead7017cb31c7cf86f9c80ff05175fa490d7c7488f1966dabd tasks/data_persona_aligned_base_50_0029/env_builder.py
@@ -412,7 +413,7 @@ fad418cecfe44395fcfd641f293f42a7d07d069862702238595527dd91c98c00 tasks/data_per
412
  94493ce13f1f732cdb9b1ce467de39300de2c4ce52ada5c5213e2c7dd5079003 tasks/data_persona_aligned_base_50_0030.yaml
413
  2d38a884e90526b3985d2dfbefc313f7a92b9e490164dbf94fdbd7117202947d tasks/data_persona_aligned_base_50_0031/_env_builder_impl.py
414
  962233981532626c1d8ad5a71a3502d076a57e494890127cdfb84172efb11fca tasks/data_persona_aligned_base_50_0031/env_builder.py
415
- 59cef9ed8531a0635ba2306939a54de0684e21ed193e5f52c0c151c475d456fe tasks/data_persona_aligned_base_50_0031/verify_workplace.py
416
  a0c6f4af4e68ff04f39239543bd0e9b952818074c5c6827a8be58e2cd7b2edb3 tasks/data_persona_aligned_base_50_0031.yaml
417
  a50b6379bd53a7c06a8de7cd843d681def34cbf074a926d8a400ea5450f5aeb2 tasks/data_persona_aligned_base_50_0032/_env_builder_impl.py
418
  0d0659c5bd27644758e79b1644369f3af8f4caf6ac7b454f1e308cd87ed98a61 tasks/data_persona_aligned_base_50_0032/env_builder.py
@@ -440,11 +441,11 @@ a9068ff071983d3dd611b2d5a2acc8617176707412c32a9815caa022b35fb4cc tasks/data_per
440
  e163a7719e34c46c247540197dc4ccb9b4ad22dd6e18dbf676817ef018e06e00 tasks/data_persona_aligned_base_50_0037.yaml
441
  0a345ac8c0b22a643e0382c660b34d0bb34f9e9c768eafd84a374a8403ca7763 tasks/data_persona_aligned_base_50_0038/_env_builder_impl.py
442
  d307444652d33c8e30fc3d2cbfa33a7be79c5f4c5205d5aad21b6a5d660f62ac tasks/data_persona_aligned_base_50_0038/env_builder.py
443
- f1d2c832a21b95ed93ded2126e4fed9c1ad7adf8f80597b91ab2297f959c047f tasks/data_persona_aligned_base_50_0038/verify_workplace.py
444
  38aae608a4c8247283e4021b8cd3d6185fa1b34f30574b23c7268ffb461a20a9 tasks/data_persona_aligned_base_50_0038.yaml
445
  f19c5b30bd5c3a9b832b2d28d05b73994e7563a8581265b337a7ca1bf54d36b5 tasks/data_persona_aligned_base_50_0039/_env_builder_impl.py
446
  f9116c57a7886975283b29c060c9989352a11d923624c433b5fb34cae2ef26bf tasks/data_persona_aligned_base_50_0039/env_builder.py
447
- fe7bcbf69164c68d78ae425711f8af3acb5e37e4e2ccc2b616794ae571b19aaa tasks/data_persona_aligned_base_50_0039/verify_workplace.py
448
  e4bb5d21f034ecfea41ce097f7f583ee9e21516288efc529f3122f1833fa7d98 tasks/data_persona_aligned_base_50_0039.yaml
449
  5046ee44b9f95c4bf1f48829c6fe39e9cb840fc55f2f424782c844e3c8bddbe7 tasks/data_persona_aligned_base_50_0040/_env_builder_impl.py
450
  fef2ef8050e166f29c89bd79ed479d597710754b3c31d00b7e11e4477c8dd9bb tasks/data_persona_aligned_base_50_0040/env_builder.py
@@ -464,7 +465,7 @@ a5ed75d43e1e30d662992b0193ee7eaf0489fdcf2dd1441b6fa507de13a91750 tasks/data_per
464
  bd8ff94edccc5a0673c4420fadd0b5b2ed0f97dffe2bc84678220bf7ed914a1e tasks/data_persona_aligned_base_50_0043.yaml
465
  250b4807517b0b3bb86ac7d661dab733e27adb4928ca59e3d8e784613014d967 tasks/data_persona_aligned_base_50_0044/_env_builder_impl.py
466
  974c8fdcd0dc2873aeb9362c3ef9aec854580103db90148adb7e1002a0c4fa45 tasks/data_persona_aligned_base_50_0044/env_builder.py
467
- 80a91bc78ea39b9d25e77fcfd7457e9dbf60a6cd02ea4fde5ab74a080ecc44bc tasks/data_persona_aligned_base_50_0044/verify_workplace.py
468
  91ec50ad850b6f8c5b9dcd6d9d6b697dbd4de208694545a4e4e4c1986d12107c tasks/data_persona_aligned_base_50_0044.yaml
469
  22503e360c49920cec2452ecb6ee9781275a8f7757bfb74f98d52b4621b2a8b1 tasks/data_persona_aligned_base_50_0045/_env_builder_impl.py
470
  ca3933e0a647e00517baf586b4608e42df835592e854e6ff03710c594f19e0f5 tasks/data_persona_aligned_base_50_0045/env_builder.py
@@ -488,7 +489,7 @@ c53229f72d98969364a510c422ad052f8831c8520aefbcc2610db2243695934f tasks/data_per
488
  33e33f82ff35a6ad6cb1092e5e36b3f5939a8efe3f084955c9086b7508fcb579 tasks/data_persona_aligned_base_50_0049.yaml
489
  621b6ad772a34c7de0253650134ff3ffa2965558360f7d3064aecec375da447e tasks/data_persona_aligned_base_50_0050/_env_builder_impl.py
490
  db6022647edb92f6923877f378d15eb624578baef866632d74b05166a7cd908b tasks/data_persona_aligned_base_50_0050/env_builder.py
491
- b6bc15bbdd9dbaa888ebe6358bcff82b42a33d149f05c0f5f365c7373848c15c tasks/data_persona_aligned_base_50_0050/verify_workplace.py
492
  39985f7f65d42723fad48a0f4f9485351aeb9d3fdba74d11987c403a5a2a5c30 tasks/data_persona_aligned_base_50_0050.yaml
493
  9892213853ca4c0261fd68c4cb4e247eaa0b21430d22b5209ea1cb29b1f4edfc tasks/data_persona_aligned_hard_50_0001/_env_builder_impl.py
494
  3e61d1d76042916afa0bd62010e86d59370e0d27c26971e7e9f3905d86b78886 tasks/data_persona_aligned_hard_50_0001/env_builder.py
@@ -500,7 +501,7 @@ e8dc8cd5f2a5f0dcacb08f6b59c79625e767d1ad33b2c98debbcdcdf481f6bd8 tasks/data_per
500
  b14d812ae12f36674a7df36748145b7beaa7806f8e15ab0bedfe6a982e8ac1cc tasks/data_persona_aligned_hard_50_0002.yaml
501
  3663abd9569ead92bf55f91bc1afa38f9b3cb9983f1878204bf36b10bb755a1e tasks/data_persona_aligned_hard_50_0003/_env_builder_impl.py
502
  0565d7f925923ff2026cf8f268ca40a8e3cb64f63d0465121b5d9ba1d8743c8a tasks/data_persona_aligned_hard_50_0003/env_builder.py
503
- e41fef1b657b2952adc0bdff2f90643d013b1184b94bb6ed362cc4dd54a980ad tasks/data_persona_aligned_hard_50_0003/verify_workplace.py
504
  88ff6a2515b4135b23b5f0d99c444f0e7bc6752d16d17ed1c1679b6c55a40906 tasks/data_persona_aligned_hard_50_0003.yaml
505
  f4d62cc9d321d62f6a3dd231a882bfc6731bb6cc8ee67be55dbc97c0307ec002 tasks/data_persona_aligned_hard_50_0004/_env_builder_impl.py
506
  c3b2349fa5381acac97516862d5853aa61853adbe3dd0f288c5b78fa3005fcb4 tasks/data_persona_aligned_hard_50_0004/env_builder.py
@@ -508,7 +509,7 @@ c3b2349fa5381acac97516862d5853aa61853adbe3dd0f288c5b78fa3005fcb4 tasks/data_per
508
  8bfdaca55308c27902185f964ebea07c47516010a236af3b7bbea5c65b394133 tasks/data_persona_aligned_hard_50_0004.yaml
509
  9377b5cddccdc96cffc589184fadc5dc44828a84a89d023a44ec86bc4478745f tasks/data_persona_aligned_hard_50_0005/_env_builder_impl.py
510
  ec61a256c79a4cf56633c6cd1e401203c96ef27ba8686fc8ecae56a2a3225a45 tasks/data_persona_aligned_hard_50_0005/env_builder.py
511
- 087029db7a27a634ae228d64507ddf2372dfd7514f5a81e7db0ec842fca2a91b tasks/data_persona_aligned_hard_50_0005/verify_workplace.py
512
  ef459e1e02cb3ed81b88ec63e8e574199b4d0aead843d3627f9d634effa39157 tasks/data_persona_aligned_hard_50_0005.yaml
513
  1b8d9ed1662013c6063095a9dee57f4eb45e90e52aaccc23665e686d7d6ee285 tasks/data_persona_aligned_hard_50_0006/_env_builder_impl.py
514
  4de8a87722a1a52230c66a4f518b0f2dc71e9dcab64c0d4ce2a5928f3bf408ba tasks/data_persona_aligned_hard_50_0006/env_builder.py
@@ -520,7 +521,7 @@ f4d5839a43ee22c0ccea148e52ffc2a668b121b954b8778cabab7b2647033caf tasks/data_per
520
  93ed4d1565625e3807ab2bd958b08f13eefdff0177fd2a3976191dac1559c5ff tasks/data_persona_aligned_hard_50_0007.yaml
521
  26e506a445f77a0753ad8925746c99bdc9504359ba99d6496a7cd39c1303f7e2 tasks/data_persona_aligned_hard_50_0008/_env_builder_impl.py
522
  342c106b617b8540cacc25ce925991e51f53b3cf7f8466705867de9aa8813c48 tasks/data_persona_aligned_hard_50_0008/env_builder.py
523
- de2dc9f4f99fbd57d7296b033e941bad29bcfc4ecf3c2348fe34d9d0df734da4 tasks/data_persona_aligned_hard_50_0008/verify_workplace.py
524
  9272346004462fd9b33b9ce938153b90b4c3ba63a20a4b3f4a238e0ae89b8b9b tasks/data_persona_aligned_hard_50_0008.yaml
525
  81d367d74e1ab93dc2c40117acf662ee1df3e59d97836a40e7694d919ae85258 tasks/data_persona_aligned_hard_50_0009/_env_builder_impl.py
526
  4c89bb7b35cbcede49935ea0a78d6ee5adf3225738523fbfe3d0161328ba05d1 tasks/data_persona_aligned_hard_50_0009/env_builder.py
@@ -528,7 +529,7 @@ de2dc9f4f99fbd57d7296b033e941bad29bcfc4ecf3c2348fe34d9d0df734da4 tasks/data_per
528
  0e7bc5aa1ee8327bd92ac848ac58551cef3542910c1f8816761cf79dc307d456 tasks/data_persona_aligned_hard_50_0009.yaml
529
  aaa724cf61cbf07ae648089a1c790880da4969085fa5b6fc1b49b42134a23b66 tasks/data_persona_aligned_hard_50_0010/_env_builder_impl.py
530
  a0b86d0d17662a9c17cbd783917140943b55b5b4264df71528558781dfd36ce6 tasks/data_persona_aligned_hard_50_0010/env_builder.py
531
- 87112b13fae9a46aae422580eccbf74f6d97f5dd69c776c3d3b7ddb103b91193 tasks/data_persona_aligned_hard_50_0010/verify_workplace.py
532
  c1a09ff9086f6ae5a6150e807b971374dde5d1f90b998701c0b189fb4732b446 tasks/data_persona_aligned_hard_50_0010.yaml
533
  06d42d54408f75e93d1b8be591f6bf259037caae86064f98f335992f619d6c3c tasks/data_persona_aligned_hard_50_0011/_env_builder_impl.py
534
  316d452fdba8ec3f9f774b94f8a2ea6d898d280b133dff09efa1c6cc55e61cda tasks/data_persona_aligned_hard_50_0011/env_builder.py
@@ -560,7 +561,7 @@ ddbca9bd917f791418d49f0086bc60885e990284ea22c94e73bce7fd8118cb04 tasks/data_per
560
  d04f43e4c15c4c935c9586b61e948637ce13668e1fc131f3ec326193df779e40 tasks/data_persona_aligned_hard_50_0017.yaml
561
  17fcd2a10401eb20c0778932a134f08aa739d51566965e4d35ba01be72a0f734 tasks/data_persona_aligned_hard_50_0018/_env_builder_impl.py
562
  98944163bb3ed53c6d7850cd088c09f15b9a634c1b03e4b0ccabb73fc7fe07c9 tasks/data_persona_aligned_hard_50_0018/env_builder.py
563
- 8c39a9476c58ff7aedad44f6be1dd9240023924389b72aaa4dca5b0b45aebe7b tasks/data_persona_aligned_hard_50_0018/verify_workplace.py
564
  5a374e07aa2bbc2ecd36b7f8706e293b1552920462256a87b4cfbc717d6e12b6 tasks/data_persona_aligned_hard_50_0018.yaml
565
  91f2a954bcc89ee384a12b36a45675d6d1af572f8c88d1ca0a534914592bff70 tasks/data_persona_aligned_hard_50_0019/_env_builder_impl.py
566
  765b8babb18dc5d8b9c7abc5cf3c8b527c8b3dc0718bc58486e177d68a812f70 tasks/data_persona_aligned_hard_50_0019/env_builder.py
@@ -588,11 +589,11 @@ aa92ca12d84e09a0364123092ba0ca97dd166c00aaf90d019d1a80696e7362b7 tasks/data_per
588
  6abe823a388bb7d9b4178bfe63c1364714a326172c110b39feaab026ca64665d tasks/data_persona_aligned_hard_50_0024.yaml
589
  10e5ab67e724c6eff45567a16296e426583b4a01191c5c9657c4f053ffc9672e tasks/data_persona_aligned_hard_50_0025/_env_builder_impl.py
590
  9623bfa7c6debc23c10fd65680189bc336c74820b51edf66d13a3256a345c3f4 tasks/data_persona_aligned_hard_50_0025/env_builder.py
591
- 75819647a143ebed4e408dd4bfe3d32a33d25ed7cd5fc7fffc444c7657f57b07 tasks/data_persona_aligned_hard_50_0025/verify_workplace.py
592
  aa6e710116826c1faec4c6db3ade51b804c0ed28f3d1786c25c6c9055ba297cd tasks/data_persona_aligned_hard_50_0025.yaml
593
  94dce0fdb078c2c0978b2d755a1ed677f12ad86f1b01eceb4a99bb2f77cbf766 tasks/data_persona_aligned_hard_50_0026/_env_builder_impl.py
594
  b5a5760af5aa6213db8ae77365d5a324260fbeb05ac38f6183bed96d2497e3a4 tasks/data_persona_aligned_hard_50_0026/env_builder.py
595
- 412a1bf5111867525e7cb5058fbc43408f5f6002048e5c3e102eab94a64c5c67 tasks/data_persona_aligned_hard_50_0026/verify_workplace.py
596
  94d37e64629a0afc19595978c77be604fd0f69e8d021eecebd606f8697e37e44 tasks/data_persona_aligned_hard_50_0026.yaml
597
  6d71034b3da4d856404318ae3e8c6964039c746d31568bd001120f9f2ee323d8 tasks/data_persona_aligned_hard_50_0027/_env_builder_impl.py
598
  e916a151a5dfc1216d3126e1fdc9d3e49d4ff21dfd0d8e1a69c481b99e652844 tasks/data_persona_aligned_hard_50_0027/env_builder.py
@@ -604,7 +605,7 @@ d22ec93e6890578d6cad793b3f75be07274ecaa2dca93d0d00ef7b075960a0c0 tasks/data_per
604
  e9482f5c8e5e45ac6ce070706b4b16229f5411a3c59c9d266d20dcd4dd40a7bc tasks/data_persona_aligned_hard_50_0028.yaml
605
  641130cbb094d46c4f6a6453bf48c4da4223b1c6b2f11378cea254a3d1550cd2 tasks/data_persona_aligned_hard_50_0029/_env_builder_impl.py
606
  bedb04030f528ca6141a3a8d98870e6298ece7296c1e9427c6ef24f155292030 tasks/data_persona_aligned_hard_50_0029/env_builder.py
607
- e11c99d8f991b3b9daa419e47d6379d3af911f321b3dc1173d4b99862f6daa5a tasks/data_persona_aligned_hard_50_0029/verify_workplace.py
608
  cb3cf4435c2d8f1b84430fba628f17a187d07500152b6869ef58ba690486596e tasks/data_persona_aligned_hard_50_0029.yaml
609
  57ec090bb1439b10bebbe4ba8effc0dd0251169a5c7afe6da44097428359edbc tasks/data_persona_aligned_hard_50_0030/_env_builder_impl.py
610
  4c060230b3d775a246e7917095863cff5352ecf49aa0861d6ad5e4b6468337e5 tasks/data_persona_aligned_hard_50_0030/env_builder.py
@@ -612,7 +613,7 @@ b64b6b1d5dfdf584acff4d6519b03cc10dda92a282fabcc853b7715780f64fe3 tasks/data_per
612
  653f70b05c6670ce5a312a40ddf98974ffdbeb079b10a62d5755dc7da3b4b47e tasks/data_persona_aligned_hard_50_0030.yaml
613
  8a7abefe1106481b16b296adad117e7c789d97297a674e8b734d8079718d5adc tasks/data_persona_aligned_hard_50_0031/_env_builder_impl.py
614
  1208d3209535d6a643c81fe45d9b2b2399e2c32b762ab32319642b12ee9d5040 tasks/data_persona_aligned_hard_50_0031/env_builder.py
615
- 59cef9ed8531a0635ba2306939a54de0684e21ed193e5f52c0c151c475d456fe tasks/data_persona_aligned_hard_50_0031/verify_workplace.py
616
  1b61c0f4805f412a3286e128d8307900ac900e769e4e146df828df3f9be32932 tasks/data_persona_aligned_hard_50_0031.yaml
617
  72c58b2d69f62db6115a3b44e28dc24b6a8478a8c2a50d78e413fbe588b2338d tasks/data_persona_aligned_hard_50_0032/_env_builder_impl.py
618
  462064396cd8a259d88b6dbdc0be51924d3119170343c1871e159d51b9bab1e2 tasks/data_persona_aligned_hard_50_0032/env_builder.py
@@ -644,7 +645,7 @@ f1d2c832a21b95ed93ded2126e4fed9c1ad7adf8f80597b91ab2297f959c047f tasks/data_per
644
  d6f4dac47bcfa65828f04f5b912efe77dda129fc70f4772512092faf7b764db9 tasks/data_persona_aligned_hard_50_0038.yaml
645
  cee58c8a384c84e40e1dafdc1dd9bf1b8a8cb45bb405b164856e84d10b164b59 tasks/data_persona_aligned_hard_50_0039/_env_builder_impl.py
646
  736d65347297195a85d31a4601ad05f4f2b5798bdc24708655c97b8055c34ab1 tasks/data_persona_aligned_hard_50_0039/env_builder.py
647
- a9480f599f21eea1eefa0d41213ebc86ceeca17afd0703ac4bfbfff3b312eb52 tasks/data_persona_aligned_hard_50_0039/verify_workplace.py
648
  03a305940543100215db9305eea51043bab58f331ff3c821170f90ceabed5f70 tasks/data_persona_aligned_hard_50_0039.yaml
649
  cccb54a1d4689bc2d39a5d82f25367675db0b86f9081f6e754b69ce8e6793885 tasks/data_persona_aligned_hard_50_0040/_env_builder_impl.py
650
  4aeaecc382f24f28cbb616749ecc37a92c699f6bf79e38a2a46c518a1f20d972 tasks/data_persona_aligned_hard_50_0040/env_builder.py
@@ -652,11 +653,11 @@ cccb54a1d4689bc2d39a5d82f25367675db0b86f9081f6e754b69ce8e6793885 tasks/data_per
652
  b93dce571e0ac493062d2fad51245f215768a562db36abe2bb8e50ca2b80e3ae tasks/data_persona_aligned_hard_50_0040.yaml
653
  368d0e34bf22c88838c5d021770c32be32410d872885d1edab41fa9d627de4da tasks/data_persona_aligned_hard_50_0041/_env_builder_impl.py
654
  7cd26a3a972cd04ff8f87382186ef63e0847a35f52c8f96098480dca21e66068 tasks/data_persona_aligned_hard_50_0041/env_builder.py
655
- 255c3f1d5428a6f22e64f3b6fbc5cf570632748982f8baacbfcd215741bcbeb1 tasks/data_persona_aligned_hard_50_0041/verify_workplace.py
656
  1eacd7393d74e2a684d26214b52dc96e71f2d5ec202738e725b33c362f42c048 tasks/data_persona_aligned_hard_50_0041.yaml
657
  fe55b1d0bfd298402f6a6dc4f979acff5fe3510cde4743ad3a7884f7fdf90bee tasks/data_persona_aligned_hard_50_0042/_env_builder_impl.py
658
  5fd181c6b52c2b26963900fcfb2a6636664a56e8adce3e8bf70797cc0b119d47 tasks/data_persona_aligned_hard_50_0042/env_builder.py
659
- c01bf548db82441bdaf99efc43738cf0bdab4f382e7b593915156e3919799be5 tasks/data_persona_aligned_hard_50_0042/verify_workplace.py
660
  c61d5a1a8a457e950266bbe86a5dc1e601baa0c0f63eed573679581b0f6a3ed1 tasks/data_persona_aligned_hard_50_0042.yaml
661
  ddd0c64952ad98b7d527bab20b85db3c2f403102ae5fb1e687f2aeab46dfd24b tasks/data_persona_aligned_hard_50_0043/_env_builder_impl.py
662
  74c8965b40c3a1d0903a6084542ba7e77fecbc8c4a9cd0b1a91cbe65dcfb7b2a tasks/data_persona_aligned_hard_50_0043/env_builder.py
@@ -664,11 +665,11 @@ ddd0c64952ad98b7d527bab20b85db3c2f403102ae5fb1e687f2aeab46dfd24b tasks/data_per
664
  9acfdb19b5742bd89764f244a16d4285f6c534c363cf619743aa7cc77cd2243d tasks/data_persona_aligned_hard_50_0043.yaml
665
  cba1af8dcbb9be4e8ce767e2bf32bd4712022eac1a08d45cec3a77cdfcce5db1 tasks/data_persona_aligned_hard_50_0044/_env_builder_impl.py
666
  9f2e91e8a8c10182bbad2884c67327bc1d05b0bfcd774167c2b9eb30839dc30c tasks/data_persona_aligned_hard_50_0044/env_builder.py
667
- abfe8c29f9e0673da311c9dc0c170302c4b2c1e5079b250650d41d4f52ea06fa tasks/data_persona_aligned_hard_50_0044/verify_workplace.py
668
  ce6178011783f0c1376190f58110c5d5243259bde9b0335b85b14423b32fe60e tasks/data_persona_aligned_hard_50_0044.yaml
669
  2f7c3b1ba5149edfa9f168c3ad807326df9f8edc800c9b39e2a7d8c0cf61ae67 tasks/data_persona_aligned_hard_50_0045/_env_builder_impl.py
670
  d182c408e7279f400601c244e2f1c7a71801ee692e981dd0e377321386e4bd49 tasks/data_persona_aligned_hard_50_0045/env_builder.py
671
- 070516b9309bf8ac0ba56bb7816c92b82b6551d131902b1c5b23c58709b1fa24 tasks/data_persona_aligned_hard_50_0045/verify_workplace.py
672
  355d9686d9318ca38b287079a45d4a50a714b19e744ec1bcc49517f35c07af68 tasks/data_persona_aligned_hard_50_0045.yaml
673
  d31aefed27596025b838d8173ee37ea7878873a4503fe42014cccbd2f60fe200 tasks/data_persona_aligned_hard_50_0046/_env_builder_impl.py
674
  9843e2b501ad3be64b7e6489b13804ff5bf44de0755d4780540bde37516e96dc tasks/data_persona_aligned_hard_50_0046/env_builder.py
@@ -684,223 +685,223 @@ f342edee864132c0e2defe6a8578a2de285e792382c8baa21054deec5c6638cb tasks/data_per
684
  00c24ba621546cb257268342831b204cabb8062991524828b527fcbc886abf62 tasks/data_persona_aligned_hard_50_0048.yaml
685
  bb6cc3f6e6c6aa37a8f7befe92c1610110b73cee53fc880c83bf14cae5f787b3 tasks/data_persona_aligned_hard_50_0049/_env_builder_impl.py
686
  2d3389cd12b5cac19e2870170bfc7cd0262f8cd7d3992f9fe206933b822767c9 tasks/data_persona_aligned_hard_50_0049/env_builder.py
687
- 48b28e1dcf1fae89659a53c448d47dcd39903bce213e818eddc4f11e5db62a90 tasks/data_persona_aligned_hard_50_0049/verify_workplace.py
688
  8c6f40192de673eda70c90d6d49289ab51f85284861eb2a90063c50713ba26eb tasks/data_persona_aligned_hard_50_0049.yaml
689
  d131ba8ad15605f073159e5cfa8f59d1f61441775e05ed0b0d3fbaddfe22e956 tasks/data_persona_aligned_hard_50_0050/_env_builder_impl.py
690
  962bc6d415f84f358a93f98c58566b35d4583c848105e62385653afc8ca4a050 tasks/data_persona_aligned_hard_50_0050/env_builder.py
691
- 6e60424e15a09625aa2cb915e82eac189a7334d598b6deab00549c5283579de0 tasks/data_persona_aligned_hard_50_0050/verify_workplace.py
692
  f246d3283b5505e1dc14898df39dd637dc3f463149c433f47df715f66e462b92 tasks/data_persona_aligned_hard_50_0050.yaml
693
  e0870061523156dcf50b3d061b41eb0b55f4f1cd500110c6baa21cc8b6c7d62f tasks/data_persona_aligned_multi_turn_50_0001/_env_builder_impl.py
694
  aa5693f630e9cb100b535d7fa4a218b909ab9ae365224b51a708dc2cbf566fff tasks/data_persona_aligned_multi_turn_50_0001/env_builder.py
695
- 0d18b9055e349452dc45fed2d38b0ad9ea4723738988381b60c4383459e344ff tasks/data_persona_aligned_multi_turn_50_0001/verify_workplace.py
696
  333636fb22be35889c8dd4548a374dc11246e3a3fc0d9a3267fbf49fb2bb5966 tasks/data_persona_aligned_multi_turn_50_0001.yaml
697
  94846961de34d7ac9d141af76fde4d100e8d7fc65a079d00104365d76bf086ab tasks/data_persona_aligned_multi_turn_50_0002/_env_builder_impl.py
698
  a7ce68a93347109f983ed0429b28ab540ea0517d3c20dc44e29a143fca98cac3 tasks/data_persona_aligned_multi_turn_50_0002/env_builder.py
699
- 052411e890e3277b5ae31abd2f5d3c970b929ebbe3f1e87cf5787db09afbfa27 tasks/data_persona_aligned_multi_turn_50_0002/verify_workplace.py
700
  e81caf1948c591f769bb11aa1910b78f8d01f2dc9c886dc1673787f01dcb1f95 tasks/data_persona_aligned_multi_turn_50_0002.yaml
701
  56b62385dc8409dc5fa0a18033446b9ecc250f93dba955aa803e0bf174f499cd tasks/data_persona_aligned_multi_turn_50_0003/_env_builder_impl.py
702
  005b608483709080698b9054c47786e7efe90831cf88c98caf750fdfb8a9006c tasks/data_persona_aligned_multi_turn_50_0003/env_builder.py
703
- e41fef1b657b2952adc0bdff2f90643d013b1184b94bb6ed362cc4dd54a980ad tasks/data_persona_aligned_multi_turn_50_0003/verify_workplace.py
704
  b27b9f0128955399af5ac73b2ffbc926839ecdca9cec6be21d9a93f4f3b7892e tasks/data_persona_aligned_multi_turn_50_0003.yaml
705
  8e39e4e7954d29970b56ec7099c3d60fb38884945628454327845c041301d57d tasks/data_persona_aligned_multi_turn_50_0004/_env_builder_impl.py
706
  daa0e5221bcb2406668d13596243e86754b65490b3ac430bc5fe0e02b2772d4b tasks/data_persona_aligned_multi_turn_50_0004/env_builder.py
707
- db2ec867511c459d3989cee2fd2efdc24ac7f1977bd53fce78bbdadfc86735fb tasks/data_persona_aligned_multi_turn_50_0004/verify_workplace.py
708
  9c40b0609198ddabb5faf5c1438dadc8c8338e5866783b968c3c6f851ee50c94 tasks/data_persona_aligned_multi_turn_50_0004.yaml
709
  cba2b5f2d524c3ac229b241b5ecaadd9bd4851ee085061bad847f6f35b103198 tasks/data_persona_aligned_multi_turn_50_0005/_env_builder_impl.py
710
  d572d99fa9e6914cfa62bb790a95c896734f1582e65b7bca58cba7fda6d678e1 tasks/data_persona_aligned_multi_turn_50_0005/env_builder.py
711
- 087029db7a27a634ae228d64507ddf2372dfd7514f5a81e7db0ec842fca2a91b tasks/data_persona_aligned_multi_turn_50_0005/verify_workplace.py
712
  3a8b7a38d884dd884117035c4c1c0c1170ea9a22aa7635f0518bc5eea72d9bd0 tasks/data_persona_aligned_multi_turn_50_0005.yaml
713
  1dc41fe560925c3bfb6b1f833364d2b6c26714b35144d33c27ed692e2d1ee9a9 tasks/data_persona_aligned_multi_turn_50_0006/_env_builder_impl.py
714
  18bbd5412ba2f4c574161cb6c570ff8566d8d1376acdeab1fb4c0dcf4705e423 tasks/data_persona_aligned_multi_turn_50_0006/env_builder.py
715
- a640a92580297c5b8a01e7d14dc295c84c8cc62bea1d17e088b69158d5c83bf2 tasks/data_persona_aligned_multi_turn_50_0006/verify_workplace.py
716
  c5b7cde4c2402bf0c72663da6d8f792178cc729f3a2da8383202a01ef07ddb78 tasks/data_persona_aligned_multi_turn_50_0006.yaml
717
  53da96cb334974bb68f6379f6b5ae23ce42c9ca5ac8c11e0e167f9122944346b tasks/data_persona_aligned_multi_turn_50_0007/_env_builder_impl.py
718
  6a93ba55bdd0b829084ca629904551c0f79a91aa457a6042bae3d225b5f48cc0 tasks/data_persona_aligned_multi_turn_50_0007/env_builder.py
719
- dfbbbcb69c4cd982f01b32988e7457dae61c1d1f286b851473eeb0d98941ebcc tasks/data_persona_aligned_multi_turn_50_0007/verify_workplace.py
720
  2cbe9b2c88264813ea9bca905cbcf16ea7a3ca7a18a00c70ff422ec4e85cf9e2 tasks/data_persona_aligned_multi_turn_50_0007.yaml
721
  9bee8ee455207c056c55f34ead641e97a7c83d0d91e5f5ed582eb4e27bf1ba9a tasks/data_persona_aligned_multi_turn_50_0008/_env_builder_impl.py
722
  8b594c39ed78a5714a289f4478075ca59bca28dd66e8d5cf720b2e7ea2b0390c tasks/data_persona_aligned_multi_turn_50_0008/env_builder.py
723
- de2dc9f4f99fbd57d7296b033e941bad29bcfc4ecf3c2348fe34d9d0df734da4 tasks/data_persona_aligned_multi_turn_50_0008/verify_workplace.py
724
  7f4e82abd2ce0b7e78134d30112e67332d32a7f34a37f882a94f39a6f7a832a3 tasks/data_persona_aligned_multi_turn_50_0008.yaml
725
  19eb062b37e50c552972258444289b269b371cca4b707fb95d98345a27c41da5 tasks/data_persona_aligned_multi_turn_50_0009/_env_builder_impl.py
726
  3fd3bd1560b093b7795bffac6256e1378d734b34692bb6fb4155566a3fd202f7 tasks/data_persona_aligned_multi_turn_50_0009/env_builder.py
727
- c46cda8fb146fabcaeb40e30e230870d250286986f6530ef8cbc782a5910c2ba tasks/data_persona_aligned_multi_turn_50_0009/verify_workplace.py
728
  fb0862f31973d092ce9bdd17a3e44ba234b2f1cf69b25550ba600a46df727b96 tasks/data_persona_aligned_multi_turn_50_0009.yaml
729
  bba2fb8f620eaeaa151bc313d9e61ad95411dcf4ea11fa837626dc0b6fc1e0b0 tasks/data_persona_aligned_multi_turn_50_0010/_env_builder_impl.py
730
  56b05bc689641318b690c0425450f367c608b35841f7f5b98da729b75d73cc4d tasks/data_persona_aligned_multi_turn_50_0010/env_builder.py
731
- 87112b13fae9a46aae422580eccbf74f6d97f5dd69c776c3d3b7ddb103b91193 tasks/data_persona_aligned_multi_turn_50_0010/verify_workplace.py
732
  b3ecbd7315c94c81aae322dbeaad3cb8e43a8780c3eea940fa7aeab7b1f14878 tasks/data_persona_aligned_multi_turn_50_0010.yaml
733
  8003c68cb6d637cc88ca863aed65fb344e817be2360e6324b432c73101fa5c40 tasks/data_persona_aligned_multi_turn_50_0011/_env_builder_impl.py
734
  b4625423f6ce3868ae90cd07bf017dfa3d85578f2926006180e03be729d41ab6 tasks/data_persona_aligned_multi_turn_50_0011/env_builder.py
735
- a9e5f27a5c24522bdbbe7455479af578efa7961e09117b9f303e2d57bde767a9 tasks/data_persona_aligned_multi_turn_50_0011/verify_workplace.py
736
  3a09c0e7a11d6474f6072bc31c2982655c55896bee8499733c8f32f441544719 tasks/data_persona_aligned_multi_turn_50_0011.yaml
737
  e4b00c3dc279210b49d7ff3eebde96ed00f689d4cbbe0f9f1db8f8967ef69549 tasks/data_persona_aligned_multi_turn_50_0012/_env_builder_impl.py
738
  0ed06754dc947f478769feedf9ffc69b231b19bb3930e80913cc92952307cc87 tasks/data_persona_aligned_multi_turn_50_0012/env_builder.py
739
- ce2498b965464208fe24275f94f3ddfad42060382409a130a931bfdf32044e1c tasks/data_persona_aligned_multi_turn_50_0012/verify_workplace.py
740
  682f72ce4ba4ac87b0b7f202d6dc4f2f5c855843264299130cbfbb8ab8ccc651 tasks/data_persona_aligned_multi_turn_50_0012.yaml
741
  220477253bce809ca1035473da392466a5009633b9312964979f664c7318861d tasks/data_persona_aligned_multi_turn_50_0013/_env_builder_impl.py
742
  36ee6290bea8542c8dacea5d70d9ae0ae967b353466240282629cc9dad6e11ce tasks/data_persona_aligned_multi_turn_50_0013/env_builder.py
743
- 6f3513f2782c863cde1c77b0db94908cb7b5844c68e991a2a25ef930d016c6e4 tasks/data_persona_aligned_multi_turn_50_0013/verify_workplace.py
744
  87163756033db05d116adbaf83171f3ff77e02c9cd8a0715f0587f68a7b91409 tasks/data_persona_aligned_multi_turn_50_0013.yaml
745
  60e5cd1d6b6fbcb2c01eafc5fe8bb0c59c74d9d1e32105a207191fe713f6af4a tasks/data_persona_aligned_multi_turn_50_0014/_env_builder_impl.py
746
  c42207780a8226358470b5328b1ecdd062aa2cd5877bad0de7cfa6dfcf7b44d0 tasks/data_persona_aligned_multi_turn_50_0014/env_builder.py
747
- 75985440766202afe1b9fc12d90d0a7fdcec966048b22419e167053ef9517e00 tasks/data_persona_aligned_multi_turn_50_0014/verify_workplace.py
748
  3849aca1cef3ca1d7635b8dbd325d970d1ec6207d45afeffc7ca619e4bddc0a8 tasks/data_persona_aligned_multi_turn_50_0014.yaml
749
  7c6577faff4bebc7961e19db2f249c1916f909a308f89922eea394380206e1f9 tasks/data_persona_aligned_multi_turn_50_0015/_env_builder_impl.py
750
  f7a54700886af3193ddf9859affea8860dfa76c1c39182c3a02d283d50dd18d5 tasks/data_persona_aligned_multi_turn_50_0015/env_builder.py
751
- 2ec8a8f765aaa869e1dc2ce9a7e1c8c54b30ae15a1875dc350bdee5bafa200cc tasks/data_persona_aligned_multi_turn_50_0015/verify_workplace.py
752
  a4ca347eb99244e243ca85bf6e4206760df3028349725e09cd91ca03cbed8ecf tasks/data_persona_aligned_multi_turn_50_0015.yaml
753
  09bca3252d8d83eb9f5eef55eb377a6410dd4d8867bf99cd7f067ee4d12ecaa3 tasks/data_persona_aligned_multi_turn_50_0016/_env_builder_impl.py
754
  574ef7bae7593f94697b980b097dfce54aca011d6dd12f4ec33d13833c975467 tasks/data_persona_aligned_multi_turn_50_0016/env_builder.py
755
- e274e884cae34708d31a9c6224f8979c5881661f16f8bc092d6a4decdef5c07b tasks/data_persona_aligned_multi_turn_50_0016/verify_workplace.py
756
  47f0f2a12c50e596768ba756aaef4ecc35c8a292f023c4e2b5afcdb36275551a tasks/data_persona_aligned_multi_turn_50_0016.yaml
757
  c643e2ad89f0ec3bec6e0dd36b61da9b8b7f87b3a6b09913328ab0783da9fcda tasks/data_persona_aligned_multi_turn_50_0017/_env_builder_impl.py
758
  1d98fb6b956e9fcc57cd8ac436ccb8bb72164fa6e81c68dbc7ab7e5c20e7ba59 tasks/data_persona_aligned_multi_turn_50_0017/env_builder.py
759
- a25423e31f1dcc75832dc2e686ea2c6d7e2e1137095917b9e4ffb4b161d6857c tasks/data_persona_aligned_multi_turn_50_0017/verify_workplace.py
760
  47e9b166b2b5f65d19bf778415e950e06d479ad8500b2c362870ad086bdc9ddd tasks/data_persona_aligned_multi_turn_50_0017.yaml
761
  2c734b483f1be0469fda215bed6657704fac8cd0949a58107d4f49b4f7fc5676 tasks/data_persona_aligned_multi_turn_50_0018/_env_builder_impl.py
762
  4797fb5c7d9af2c55afb658e3df552ca12d3489305802f658da44e2a8300045c tasks/data_persona_aligned_multi_turn_50_0018/env_builder.py
763
- 8c39a9476c58ff7aedad44f6be1dd9240023924389b72aaa4dca5b0b45aebe7b tasks/data_persona_aligned_multi_turn_50_0018/verify_workplace.py
764
  fb8116b17ce62d333453921784668a580f7f72e1dc8153ba39d9254bb48fc8bb tasks/data_persona_aligned_multi_turn_50_0018.yaml
765
  5c619a88b5de3cff6e4d453ca9bc26c80ab888a9074007c10fbd4e9b8e3769e2 tasks/data_persona_aligned_multi_turn_50_0019/_env_builder_impl.py
766
  a79017e9c78017b90ec33b2d8a7d56d9f5f7fd22016a4cd89755045d35c29e7f tasks/data_persona_aligned_multi_turn_50_0019/env_builder.py
767
- 14cb198c768f63d47ec0cfb1ad652a74eeada4b579287c6d6a134ae57cc6267a tasks/data_persona_aligned_multi_turn_50_0019/verify_workplace.py
768
  70d6295d4f3bfcad9088cbdf1e1950a1fa51d03b880a6d1b5c886c4f6819793e tasks/data_persona_aligned_multi_turn_50_0019.yaml
769
  c738370ce6a9a4e9ced171c1fe8cecf0990082c2787269e00699d4903c100cad tasks/data_persona_aligned_multi_turn_50_0020/_env_builder_impl.py
770
  8f313309971ecb7df8ce2132f41c2661137c308a82a62cd92840ba1b906f1270 tasks/data_persona_aligned_multi_turn_50_0020/env_builder.py
771
- 5f2c6eb21d59a317cf1b3b1e7e07b71aff33bf6e578b72cf924c6fe964ea962f tasks/data_persona_aligned_multi_turn_50_0020/verify_workplace.py
772
  85c811f41301ba28aea1c24422ef98b822d4b22a9d564bca89a6b88bcfbab2c1 tasks/data_persona_aligned_multi_turn_50_0020.yaml
773
  3be98a23963dbc1e303348567aee69aea0aee4fe028bc96f2599d5e7529f2fea tasks/data_persona_aligned_multi_turn_50_0021/_env_builder_impl.py
774
  0edfd1c1ac717a0729da0d981c82c4ffddfbc0184921e75b2fb7b16da8ea2b62 tasks/data_persona_aligned_multi_turn_50_0021/env_builder.py
775
- 933a5f52e2f3f5ee90c596c292ac5d3353e29a7682b604ab7bb5b377acd3b75b tasks/data_persona_aligned_multi_turn_50_0021/verify_workplace.py
776
  44754ccd1710c477dfca7ecea24b3f8e2dd37129598b6f855335929db6828915 tasks/data_persona_aligned_multi_turn_50_0021.yaml
777
  76314efd0543d714564316e3f39d8590f93b1644a3a2c33d57287c21f9596b37 tasks/data_persona_aligned_multi_turn_50_0022/_env_builder_impl.py
778
  703067eb33128d48bda186519dfb9166f7713746138162b0b34ad7c3a8483370 tasks/data_persona_aligned_multi_turn_50_0022/env_builder.py
779
- a41f64c73b779c1718cdc013fdbababd48d715a07a6cb1f811e9b5aae883d338 tasks/data_persona_aligned_multi_turn_50_0022/verify_workplace.py
780
  2856c77f67701b8a42c05f22160dde6701ca2be1ec31af7494056431c4c90c34 tasks/data_persona_aligned_multi_turn_50_0022.yaml
781
  d7a1f604c95694bdfc8f349a31b774f006747711a6df4db92bd3c3aaec990275 tasks/data_persona_aligned_multi_turn_50_0023/_env_builder_impl.py
782
  a09ea95b569dff0baff3c85af3192fdc8d2697beb4d7d275b1286b9a22167682 tasks/data_persona_aligned_multi_turn_50_0023/env_builder.py
783
- f8150ebd4c0d96948c9d63c0b4c35d4f76321c3451fbf46a0f858f5dd6f743d8 tasks/data_persona_aligned_multi_turn_50_0023/verify_workplace.py
784
  65ff7137784414b5135c907b66c1325b88c0bd4349868af5d69418446fdd4ff8 tasks/data_persona_aligned_multi_turn_50_0023.yaml
785
  bdcb9cfb34d07f1f005497b3e51eb8a3ec486accbb314bfa3549a37f06fb2b56 tasks/data_persona_aligned_multi_turn_50_0024/_env_builder_impl.py
786
  fbc69703bb119564f71c801242fdd7b998ddaabf708c348c39907f8e9a152f55 tasks/data_persona_aligned_multi_turn_50_0024/env_builder.py
787
- 21cc9c6b17a14453efc90a4847a0d6b398aa9af500c90323a47a077da379d719 tasks/data_persona_aligned_multi_turn_50_0024/verify_workplace.py
788
  272a826b3fd8d66d30d336080fc16fcc48895fb8bb7eb45efa013a5e893063e7 tasks/data_persona_aligned_multi_turn_50_0024.yaml
789
  2d4b64c0f1268d3990b9bd5b89b7109881b13535cd1446567f697640a908a06a tasks/data_persona_aligned_multi_turn_50_0025/_env_builder_impl.py
790
  362812e444f590b5082c96d2d2549d64a1d1bf614fa3f3b24bb91525199c3062 tasks/data_persona_aligned_multi_turn_50_0025/env_builder.py
791
- 75819647a143ebed4e408dd4bfe3d32a33d25ed7cd5fc7fffc444c7657f57b07 tasks/data_persona_aligned_multi_turn_50_0025/verify_workplace.py
792
  b1f40c5349625101cfde2b597db7d8cc34c5e42c6c3caabc902115dbd19ed6e0 tasks/data_persona_aligned_multi_turn_50_0025.yaml
793
  6c463393ff0f0e82ee82f363cf49baf979dc1fa9033b4947396b222619c347c1 tasks/data_persona_aligned_multi_turn_50_0026/_env_builder_impl.py
794
  6d252c324058ded95411c3b8413678834a2342ecde91ec1b34bd8a44fe3918a8 tasks/data_persona_aligned_multi_turn_50_0026/env_builder.py
795
- b2fc67de9d1571b98c481564aa4a9457ada54ec4e888f8241d5d40a7b631ac99 tasks/data_persona_aligned_multi_turn_50_0026/verify_workplace.py
796
  b442abf8cb9ee1102004f826e795e06bad43f06c5d7bde8027e7e9b1d78d90ff tasks/data_persona_aligned_multi_turn_50_0026.yaml
797
  c0006cbabe3fdb665f6ec3447b20731e04b54f661d374135ec1a49ac4422f76f tasks/data_persona_aligned_multi_turn_50_0027/_env_builder_impl.py
798
  d9c5cfa216c64aab5a0aaa0c743209ecbaaee7b21c208b7be1bfa5187d5c596c tasks/data_persona_aligned_multi_turn_50_0027/env_builder.py
799
- d182de48eacc413d96966204d22d3f85d5dbdf6d09d717ceb32db50076b4e2b1 tasks/data_persona_aligned_multi_turn_50_0027/verify_workplace.py
800
  44311badecf8a14f9922783009223547d2906fe3c36a785555037480e4013d00 tasks/data_persona_aligned_multi_turn_50_0027.yaml
801
  060264e3bf2aff0a3968301b7a58b5d22ac0ba2de099fd51c7bd6899399712da tasks/data_persona_aligned_multi_turn_50_0028/_env_builder_impl.py
802
  1309ac0d0d1c41bf3c49be4fbebf6498c370653415c3c04582a746e70cc95aa6 tasks/data_persona_aligned_multi_turn_50_0028/env_builder.py
803
- 1ff86299dbe258b4d5d5a6ef331188d97aeb437ec121db349e5650fbb36ce067 tasks/data_persona_aligned_multi_turn_50_0028/verify_workplace.py
804
  a3b8b7dbb01682d4e43b260fcf37e0e7667556fa29bba9e07d8544cef22800da tasks/data_persona_aligned_multi_turn_50_0028.yaml
805
  195613c7a4246e3cb5b333388a1cc4676097b9eb5ab5bdcba971acca63b3a7c0 tasks/data_persona_aligned_multi_turn_50_0029/_env_builder_impl.py
806
  97fc23c9048c1c2831a9101ed991365886af0dea36ba89bf44b69df8c9b0a413 tasks/data_persona_aligned_multi_turn_50_0029/env_builder.py
807
- e11c99d8f991b3b9daa419e47d6379d3af911f321b3dc1173d4b99862f6daa5a tasks/data_persona_aligned_multi_turn_50_0029/verify_workplace.py
808
  9218446a456d501644cd5caa96b0c7ef5b80c84baea1b12cf14f78c9bed7b5f0 tasks/data_persona_aligned_multi_turn_50_0029.yaml
809
  24c92dac93cd92431a861729d9f458e17ccc7df27598a58ba06b5add1446e4e5 tasks/data_persona_aligned_multi_turn_50_0030/_env_builder_impl.py
810
  712a694746bc6018a9c34b9c9c44ed9b47016d2a021faa98f1dc0e88ee723cfc tasks/data_persona_aligned_multi_turn_50_0030/env_builder.py
811
- 2e79ce161b1406344882d009663d8b8943d02557111738f6fa0bc8532aafe08e tasks/data_persona_aligned_multi_turn_50_0030/verify_workplace.py
812
  982f6d3ee34b2b08c5fed5b60bb4efa07276e7c672ccec891066334330e9936b tasks/data_persona_aligned_multi_turn_50_0030.yaml
813
  18e2d97bb019c59ab18a075164511385e25b922d32de046ff147ba58c4cd42b6 tasks/data_persona_aligned_multi_turn_50_0031/_env_builder_impl.py
814
  580eb15a196e6e37cc128e088dfe4ca6c1f17a6f52e42c5b60ce3a583d3f7566 tasks/data_persona_aligned_multi_turn_50_0031/env_builder.py
815
- 59cef9ed8531a0635ba2306939a54de0684e21ed193e5f52c0c151c475d456fe tasks/data_persona_aligned_multi_turn_50_0031/verify_workplace.py
816
  00acac9c0ec49309b8c53bb84f43bacb1457a35c26da794dc6d1c7e1d8b3a8b9 tasks/data_persona_aligned_multi_turn_50_0031.yaml
817
  4f7e0c6b7d0085f1e41d1b0d8d734a8bb047fd9a27013a1c98ce559c73cfaabc tasks/data_persona_aligned_multi_turn_50_0032/_env_builder_impl.py
818
  38bd56ae9faf3307e0c61e4ce5369024f9ad7988c5008c2c23817e582bba4cb2 tasks/data_persona_aligned_multi_turn_50_0032/env_builder.py
819
- de768850244d9c20a23141541493617b104424e405d9efd19835268b2bfe8ced tasks/data_persona_aligned_multi_turn_50_0032/verify_workplace.py
820
  fab672100494037ff01f33a25ece72b14d6123c7bfd3541ea7d5b85cd72d6b01 tasks/data_persona_aligned_multi_turn_50_0032.yaml
821
  f14ffd8e29bc23f7005434dd6afb2837f8a92826f9a22982cabb536489196cac tasks/data_persona_aligned_multi_turn_50_0033/_env_builder_impl.py
822
  8ac1faf9e5f109e1e6cd9ca568a343d5f4d29b865787f99ad5071eaeabce1826 tasks/data_persona_aligned_multi_turn_50_0033/env_builder.py
823
- 521bfcaebb371d477bca2cba883f2d7527ed8e8dbd1d3f8dfb5fa04c86eee3dc tasks/data_persona_aligned_multi_turn_50_0033/verify_workplace.py
824
  12711b324c0e31dced39daf7ac9addddaa7bc8a78032195b383f5477d320dfc4 tasks/data_persona_aligned_multi_turn_50_0033.yaml
825
  bbbe53f28b848f7781ad8954802120729539fb83886aaaf460e234d2fbb5a7ac tasks/data_persona_aligned_multi_turn_50_0034/_env_builder_impl.py
826
  0066b33da310f06a5a083507444277ba79af6664ecbe0f75d4d53a8e5412fc15 tasks/data_persona_aligned_multi_turn_50_0034/env_builder.py
827
- 0d200a80fa6a64d3ae9151388f3595103e43a31e800445bf04eb7ce32fc541b9 tasks/data_persona_aligned_multi_turn_50_0034/verify_workplace.py
828
  52d961bd0000bd9cc7bcbeb74e5d2c99dbe321cb0973f66553ef8f7e17386b8c tasks/data_persona_aligned_multi_turn_50_0034.yaml
829
  5ae81d9ae4a23019537057c9a04c05d30daef5fa555afeeb690620cb2e956b0a tasks/data_persona_aligned_multi_turn_50_0035/_env_builder_impl.py
830
  83b535bccea97a26cf2fbf92a5611ed09d653371a76f32dff39ca89deaff03d7 tasks/data_persona_aligned_multi_turn_50_0035/env_builder.py
831
- b7a1ccb588a8a997c4b3821b6f8c1aa34b3e166507e53294c3bc197684f0f695 tasks/data_persona_aligned_multi_turn_50_0035/verify_workplace.py
832
  5e5801d61ce1eb2ccbeb4a045e0535298ebd7ded6bd82c334326ce9a57d8739a tasks/data_persona_aligned_multi_turn_50_0035.yaml
833
  c8edcd299904fb731fa74f8edf5ab980cf20e9ab6af8b33de4651c06c8afdfb2 tasks/data_persona_aligned_multi_turn_50_0036/_env_builder_impl.py
834
  0841055ba97d87d67fdcb56825f3b9b69a720687ae728d9fc73c70d428adc32e tasks/data_persona_aligned_multi_turn_50_0036/env_builder.py
835
- 6f389d1ce610ee2b4f426647ec7000441edff1d8fc40952a994ce88b4fabaf37 tasks/data_persona_aligned_multi_turn_50_0036/verify_workplace.py
836
  2ab96542a793e709c6a314e11edc7fe6118d4669ccfd69771b5fb37d194c828f tasks/data_persona_aligned_multi_turn_50_0036.yaml
837
  8e9ae3552c7b3cdf4fd1b743169708de0015351ecad1ceeba33dc220d82f5833 tasks/data_persona_aligned_multi_turn_50_0037/_env_builder_impl.py
838
  74c8ecd7d7eb79ec1816e362a855fc03c630d9394e80d0a522f850165052b500 tasks/data_persona_aligned_multi_turn_50_0037/env_builder.py
839
- a9068ff071983d3dd611b2d5a2acc8617176707412c32a9815caa022b35fb4cc tasks/data_persona_aligned_multi_turn_50_0037/verify_workplace.py
840
  871eb76552d080398c78a7cec0e4b052d1346a812aa617496e9d28f323771f92 tasks/data_persona_aligned_multi_turn_50_0037.yaml
841
  f42b5dc110a6237bf594811fba9bbd1e11c0140b9bf7123d1bc51e980b7ed502 tasks/data_persona_aligned_multi_turn_50_0038/_env_builder_impl.py
842
  871c46ddcd3c0cc59161027154f797fce686b2f5a5c18439f2459b97d1f2353d tasks/data_persona_aligned_multi_turn_50_0038/env_builder.py
843
- f1d2c832a21b95ed93ded2126e4fed9c1ad7adf8f80597b91ab2297f959c047f tasks/data_persona_aligned_multi_turn_50_0038/verify_workplace.py
844
  7d959df4fe308e26b221121fd0515965c21baf952e6188329bcd7f154f713b33 tasks/data_persona_aligned_multi_turn_50_0038.yaml
845
  a358a8ccc640abf6f5ef8c16e4c92cefd4a4b1d60f893d4a8a8693dc21c43596 tasks/data_persona_aligned_multi_turn_50_0039/_env_builder_impl.py
846
  f03b13c0760e04f97ec5afe38099a885868cfd26d8dd4d689e797086abf6eff1 tasks/data_persona_aligned_multi_turn_50_0039/env_builder.py
847
- f3fef8bbab5eee0bb76fd54789f52c8613172d54b3cde315f4bf6dcb8269db16 tasks/data_persona_aligned_multi_turn_50_0039/verify_workplace.py
848
  5fd9a8dfa951f82891bf336194bad07cd698ac839613be821416214abdc45b35 tasks/data_persona_aligned_multi_turn_50_0039.yaml
849
  f6702cc7ca4c9a4c7766b3c19a648b0cdfb21684e0cc874e3eeba7d1b814b5cf tasks/data_persona_aligned_multi_turn_50_0040/_env_builder_impl.py
850
  40a157919ea9f49fc56c73354ec53675312fdd40a221d32cfaafeee5a4542a6d tasks/data_persona_aligned_multi_turn_50_0040/env_builder.py
851
- 144f4cd96cd73e1eb440b3e1ba5112002cbef27a3b13e8a072ad2dd1b7ac5efd tasks/data_persona_aligned_multi_turn_50_0040/verify_workplace.py
852
  88fa960bc70839336fc4f4ef6dc4be347c91806648d02987c9dc3c6960a7d62c tasks/data_persona_aligned_multi_turn_50_0040.yaml
853
  226db282102ffde0d3285e4f5dba5f12582b6cb40e7575f71fdd80e55ac0d8fb tasks/data_persona_aligned_multi_turn_50_0041/_env_builder_impl.py
854
  bd0adce0018d723a97b587c08191e4561c98807881b8de5533df0c7ac62c3689 tasks/data_persona_aligned_multi_turn_50_0041/env_builder.py
855
- 255c3f1d5428a6f22e64f3b6fbc5cf570632748982f8baacbfcd215741bcbeb1 tasks/data_persona_aligned_multi_turn_50_0041/verify_workplace.py
856
  b7830cda94a157e8ee2152545ac021b7b030b91883ff6e67ee8c32fdd1b07ece tasks/data_persona_aligned_multi_turn_50_0041.yaml
857
  b826b95b287d0c441fc2f8943f25e783203fd14d0457e93684599c507956b801 tasks/data_persona_aligned_multi_turn_50_0042/_env_builder_impl.py
858
  80754de9f54c53cee72e4121334de887ee92b79d7a967a3b86cf985e33b2f6f0 tasks/data_persona_aligned_multi_turn_50_0042/env_builder.py
859
- c01bf548db82441bdaf99efc43738cf0bdab4f382e7b593915156e3919799be5 tasks/data_persona_aligned_multi_turn_50_0042/verify_workplace.py
860
  e97fe1782faaacc41fe8709b59c71fd3e61ca30098c21ea7bef75d5319433c23 tasks/data_persona_aligned_multi_turn_50_0042.yaml
861
  bc94a1a9d8edce537f51e3dbcdb56c5cd5c44d61effd937eb4e9b142f26e539d tasks/data_persona_aligned_multi_turn_50_0043/_env_builder_impl.py
862
  4e8977b9be57fc42568674d2664e5ca27569bb49b6ea075a7cf024ea88de70dc tasks/data_persona_aligned_multi_turn_50_0043/env_builder.py
863
- 5be63014fc6175b88e5beb572c0baf78707c850dbfc5fb32ad6e402542413157 tasks/data_persona_aligned_multi_turn_50_0043/verify_workplace.py
864
  cfde3128cf17843380a7d244ba6dd2d3113594a8c35bd2a07b73716504a1e237 tasks/data_persona_aligned_multi_turn_50_0043.yaml
865
  c7846fb438cb46bafbd8b73427e89cf9ec898258802996231e06a034e80cbc62 tasks/data_persona_aligned_multi_turn_50_0044/_env_builder_impl.py
866
  fc215986f1c8e4d29d5474315f49b76e227d5aaf095bc23304bd0345574cf07e tasks/data_persona_aligned_multi_turn_50_0044/env_builder.py
867
- bd4f268ac51cf356f72ef613186a695dd78392674c433d38ba0dea8f1b3a6387 tasks/data_persona_aligned_multi_turn_50_0044/verify_workplace.py
868
  48b45df848822f760f65c743e77f425971c4af190631be7e9a218c5cc83f8f3e tasks/data_persona_aligned_multi_turn_50_0044.yaml
869
  3a4e143a8ad078eb251ca69de34ba587c270b7b6651060823cc34afe9d264e4c tasks/data_persona_aligned_multi_turn_50_0045/_env_builder_impl.py
870
  0d51165aca6030bd0d8360e05660e6a8cd2b8351cadd0a89054551b9ea0244bc tasks/data_persona_aligned_multi_turn_50_0045/env_builder.py
871
- 070516b9309bf8ac0ba56bb7816c92b82b6551d131902b1c5b23c58709b1fa24 tasks/data_persona_aligned_multi_turn_50_0045/verify_workplace.py
872
  9d6cee36d66dc5afb78db5980b1cbcec9cbbb21cb1b6597764bf7d308acd8616 tasks/data_persona_aligned_multi_turn_50_0045.yaml
873
  ff492138561fbc06ac80201ee47e004adca3cfc4187d35dabdad7a75889a0ec7 tasks/data_persona_aligned_multi_turn_50_0046/_env_builder_impl.py
874
  1c3ba59afd76b331fe62b35cb6d4a0efdd04e69e4910ebd3cf0067d4fe765b75 tasks/data_persona_aligned_multi_turn_50_0046/env_builder.py
875
- 375292151b118c6d6760982a7a5ef9b416bab56ac9572b6109bf9a21a0415ea1 tasks/data_persona_aligned_multi_turn_50_0046/verify_workplace.py
876
  dc011dd019bc9673cb5916799b91bd9c9145571f009fcf34657b2517fda83a09 tasks/data_persona_aligned_multi_turn_50_0046.yaml
877
  5b5ed722e3df8926facbc4f6620b397d8e7e3fb3a605e46877111831485d0868 tasks/data_persona_aligned_multi_turn_50_0047/_env_builder_impl.py
878
  32f976d9480f4c2775ecdcfb3b875985fbc65f3ca49337bfd43f2c4696f8258b tasks/data_persona_aligned_multi_turn_50_0047/env_builder.py
879
- 5e7889cc67c56eb055dbb79c3cd259c2d2ebdcf22c08d5eef48f104d944528f7 tasks/data_persona_aligned_multi_turn_50_0047/verify_workplace.py
880
  c3f617ca33e27dcca0fe41dcd601b8b1e5d3c11325de1ca8f31a969307d8bffd tasks/data_persona_aligned_multi_turn_50_0047.yaml
881
  09f26a5bf64800014c2f0244624998f97df1e39dbbd4c85555f6b6c8b7abda8c tasks/data_persona_aligned_multi_turn_50_0048/_env_builder_impl.py
882
  01d5378944c09344f9c3627ecfb90a633b5cdb76ff858d6470ac2c4e51414375 tasks/data_persona_aligned_multi_turn_50_0048/env_builder.py
883
- 011233eea04fecbff2ccdbb3285fe1b68c003e0093b33dd33eab30ad82cb5a90 tasks/data_persona_aligned_multi_turn_50_0048/verify_workplace.py
884
  d58c989e4fe9248e102558deb63fc15c0084fcbdd0ee4b0b01d77a0bcff78562 tasks/data_persona_aligned_multi_turn_50_0048.yaml
885
  21c8013efb0e96630bc2c90040832c130e4208462d8c0ba80119af6afacffd4a tasks/data_persona_aligned_multi_turn_50_0049/_env_builder_impl.py
886
  4fac6acb9722be1a134dd388f6a895bfb17e52244ee15d6282fa71f8253dbe6e tasks/data_persona_aligned_multi_turn_50_0049/env_builder.py
887
- 48b28e1dcf1fae89659a53c448d47dcd39903bce213e818eddc4f11e5db62a90 tasks/data_persona_aligned_multi_turn_50_0049/verify_workplace.py
888
  644ddefeb0fbee316e2f60b216062ef146f90b1a67b138051e30b005561bbced tasks/data_persona_aligned_multi_turn_50_0049.yaml
889
  b191844585401540c79229e85dcbcb61d3229d50c24f77cf22620e30c3b83ee1 tasks/data_persona_aligned_multi_turn_50_0050/_env_builder_impl.py
890
  71d53a790ea3c2a3cd3f2168b934cc8fa3d25acc49dc3831c2ab0a9da1663b1f tasks/data_persona_aligned_multi_turn_50_0050/env_builder.py
891
- fb4ee89bee4b3766f56f37688b78d49d8c6491e99bab0cb0ce2bbfb4af274f8d tasks/data_persona_aligned_multi_turn_50_0050/verify_workplace.py
892
  d4b7b8462bf456d77f9914c931edb42a1f03ca9de288a3caf7a72c2414cfcc2a tasks/data_persona_aligned_multi_turn_50_0050.yaml
893
  9ba5001c18264e7492816c699544e57c474ac01319b5199ec419f613dd5d236a tasks/data_persona_aligned_skills_50_0001/_env_builder_impl.py
894
  430f0da57468b56d4d7cd5d665c3c4233d8fd9c16ad71c99f2d44a88fc2d5b59 tasks/data_persona_aligned_skills_50_0001/env_builder.py
895
- 0d18b9055e349452dc45fed2d38b0ad9ea4723738988381b60c4383459e344ff tasks/data_persona_aligned_skills_50_0001/verify_workplace.py
896
  4af3e9203475eeea4f938b3afc4b9e720008b850b88561dcb8fee7d3bab40bfd tasks/data_persona_aligned_skills_50_0001.yaml
897
  fc24bbeda0fdfbc36b8f1af148470a2d5aa53bd49cbcde7bbc69fd3d2e662351 tasks/data_persona_aligned_skills_50_0002/_env_builder_impl.py
898
  75a883f3da5b76f3f3dc087ed2af67352131c55627aa8f1bb4bc0bcd4ecedd71 tasks/data_persona_aligned_skills_50_0002/env_builder.py
899
- 052411e890e3277b5ae31abd2f5d3c970b929ebbe3f1e87cf5787db09afbfa27 tasks/data_persona_aligned_skills_50_0002/verify_workplace.py
900
  1b0ec155c383fb6e1132bf92c5abaf53349ca5119e232db1a0abc18400936cd0 tasks/data_persona_aligned_skills_50_0002.yaml
901
  bfa32aa5d8614b836be8c879484034a8d62dad89e85578450f5adfb2a9ab88f1 tasks/data_persona_aligned_skills_50_0003/_env_builder_impl.py
902
  53d248a85bb661a1cc67b1f96fbf9ea638ab2b6556d337f23e5a432f794b62e6 tasks/data_persona_aligned_skills_50_0003/env_builder.py
903
- e41fef1b657b2952adc0bdff2f90643d013b1184b94bb6ed362cc4dd54a980ad tasks/data_persona_aligned_skills_50_0003/verify_workplace.py
904
  885f054d78faf097189ee04b97fe326a7944218babce1175ec330352de7dde24 tasks/data_persona_aligned_skills_50_0003.yaml
905
  d104f49cecdae66d6a664bc99627dda16f416a6864b398dc224d76ccffde78ea tasks/data_persona_aligned_skills_50_0004/_env_builder_impl.py
906
  0d3147dfb30a7098c44f12446ea2fa2da39af6b68c9d00775b28cb2d9217cd49 tasks/data_persona_aligned_skills_50_0004/env_builder.py
@@ -908,7 +909,7 @@ e8fa3988bfebfa0935e39b61823b5d549ae4831df5a96bb0352219d5cbf89525 tasks/data_per
908
  2e41276eb8dec0024c82210103919dfc64880d3af4af90a10172a2aeede57072 tasks/data_persona_aligned_skills_50_0004.yaml
909
  f0265e16e5276519ba1562ef3c356132b5cd8cbdae845e7c0423c8fdcb80e7b7 tasks/data_persona_aligned_skills_50_0005/_env_builder_impl.py
910
  5ea44bc46399d477cc8ca7bc611eed0992cb0ff05fe4cc40e175362389ae41db tasks/data_persona_aligned_skills_50_0005/env_builder.py
911
- 087029db7a27a634ae228d64507ddf2372dfd7514f5a81e7db0ec842fca2a91b tasks/data_persona_aligned_skills_50_0005/verify_workplace.py
912
  b63105179c6ed7ee525a902045ac208c3f9a51df503ee9eaf708429bfb4c742b tasks/data_persona_aligned_skills_50_0005.yaml
913
  6c24e33be7b534bc2f825b514b962335d5dee4ca6b8329af9da055993bf92072 tasks/data_persona_aligned_skills_50_0006/_env_builder_impl.py
914
  344f2bc4bfcc1b8231fe15c7ff2b5a30f343ad02527c6c90f02904f8fd027dab tasks/data_persona_aligned_skills_50_0006/env_builder.py
@@ -932,7 +933,7 @@ a3da02eda2b217178d15b7111cf07592e6597185b66b338b6720f43c9eb5d659 tasks/data_per
932
  ce5af4d9f8f8cce1cffac88c30be0075ced9d39e3288f35572041a1172a12588 tasks/data_persona_aligned_skills_50_0010.yaml
933
  7c825463e343c1f07c463fb265309a6ee73fa3c2868230a41072e806a0d386c5 tasks/data_persona_aligned_skills_50_0011/_env_builder_impl.py
934
  737637e643d3c9e11ce2de1b00d0d643947020efdc8f217281f67acef6ddfebd tasks/data_persona_aligned_skills_50_0011/env_builder.py
935
- a9e5f27a5c24522bdbbe7455479af578efa7961e09117b9f303e2d57bde767a9 tasks/data_persona_aligned_skills_50_0011/verify_workplace.py
936
  fe955467d70c0c91443b7cbbf1af4652139c58edeee259ae5a0d1a57ff1b1d34 tasks/data_persona_aligned_skills_50_0011.yaml
937
  2e4c21330127825cd38bb129db319ad3da984675957158ea76755166b6d69dc6 tasks/data_persona_aligned_skills_50_0012/_env_builder_impl.py
938
  e179c7026efddd0374bc6f841d4ba8bf44e809af7feab3e5f8a52db7f64676d1 tasks/data_persona_aligned_skills_50_0012/env_builder.py
@@ -944,15 +945,15 @@ efbff1fb446a71d149c262bc86542c0b0e81ed3c53b9d87d344d4f47186ceaa3 tasks/data_per
944
  08b46e279e9768961e7099589b3fabfcd49f9f68f764234ba2426bb4a8c75944 tasks/data_persona_aligned_skills_50_0013.yaml
945
  8c94e13b8c2e270164732c96ed73b108dc961da4c6d2c7e02c7b2807935960f9 tasks/data_persona_aligned_skills_50_0014/_env_builder_impl.py
946
  c736803229566964a42a2eb17ec13d79bd106447f1a31890cbe22a761df25e65 tasks/data_persona_aligned_skills_50_0014/env_builder.py
947
- 75985440766202afe1b9fc12d90d0a7fdcec966048b22419e167053ef9517e00 tasks/data_persona_aligned_skills_50_0014/verify_workplace.py
948
  4f04eb44f689489ea3a7733917c527d679adad8e11dcab3e1449c016b764b17c tasks/data_persona_aligned_skills_50_0014.yaml
949
  354f87b22d9189ad51894259d7ff75bb026d010a72e660584bb326f004082ad8 tasks/data_persona_aligned_skills_50_0015/_env_builder_impl.py
950
  aebcca1c91e3c94958570841f26677ce36520b1fe4245cb02bd50f46576a81c2 tasks/data_persona_aligned_skills_50_0015/env_builder.py
951
- 2ec8a8f765aaa869e1dc2ce9a7e1c8c54b30ae15a1875dc350bdee5bafa200cc tasks/data_persona_aligned_skills_50_0015/verify_workplace.py
952
  7a27461aa88ba0eee05a784d2a47012e839b3e5ec9c5e4f111aaad6ef7b20123 tasks/data_persona_aligned_skills_50_0015.yaml
953
  cc779485fe3eb4f49c1e7f813deb9296b8e9d859ae5385dc74dc713b637a458a tasks/data_persona_aligned_skills_50_0016/_env_builder_impl.py
954
  c333538022d2e41613ae1bf79dae2917581c5d09a19f38fcb2f20781844b6281 tasks/data_persona_aligned_skills_50_0016/env_builder.py
955
- e274e884cae34708d31a9c6224f8979c5881661f16f8bc092d6a4decdef5c07b tasks/data_persona_aligned_skills_50_0016/verify_workplace.py
956
  fe8153cfb7818ca1a90518bbf1d7e87f615e3b7fb57b4c0d1a188c94fb0b800a tasks/data_persona_aligned_skills_50_0016.yaml
957
  b9911a24f4c2daacc3607ed812e58abc927d8b3c1b0b32ad82294352141e947e tasks/data_persona_aligned_skills_50_0017/_env_builder_impl.py
958
  60f0d4e7033f8216a46abf7cd841b23b2222528c40857630c868a17f58df5fcd tasks/data_persona_aligned_skills_50_0017/env_builder.py
@@ -960,7 +961,7 @@ b9911a24f4c2daacc3607ed812e58abc927d8b3c1b0b32ad82294352141e947e tasks/data_per
960
  8326b7b45697e7c849bcf5b52a1aef8efe8c82b7d8be56151b08d4d002a67f31 tasks/data_persona_aligned_skills_50_0017.yaml
961
  521c60bbf958efb0e7ea8b38c32d2f617ecad0f09b460cc8313dec307499fb3e tasks/data_persona_aligned_skills_50_0018/_env_builder_impl.py
962
  d731a7ec2ab7ccf43db400b90e85646c3342a2a614adcdc225fe8c5cd7bcc50a tasks/data_persona_aligned_skills_50_0018/env_builder.py
963
- 8c39a9476c58ff7aedad44f6be1dd9240023924389b72aaa4dca5b0b45aebe7b tasks/data_persona_aligned_skills_50_0018/verify_workplace.py
964
  e873b6c140b7b534715ca2ceac136fe62591be2cc40503e139d504b9e82b02b0 tasks/data_persona_aligned_skills_50_0018.yaml
965
  d9805c03d587b088b94ccccae6560b01fa66433ca2e441276e091887340f4015 tasks/data_persona_aligned_skills_50_0019/_env_builder_impl.py
966
  fa6755f08d5e6f0d9d4f35271d7fe03042787656ac4bbbfe3444c505905473f2 tasks/data_persona_aligned_skills_50_0019/env_builder.py
@@ -972,11 +973,11 @@ ee77c6e74bfd8d74c5cbc89a3b5a3e00bb3ad9df175af90a343689784581400b tasks/data_per
972
  fabd3cf3944965bb4f516774215504241e9e3db295e73a5ceb4586668d595bba tasks/data_persona_aligned_skills_50_0020.yaml
973
  ed9434d766f1a2e81a0daae2cfdb59820535b44e852ef1dec0d051869ae4cf59 tasks/data_persona_aligned_skills_50_0021/_env_builder_impl.py
974
  1521cab300a19371492873f2bb15ed83a2782e98109cb5e89d04043820b50694 tasks/data_persona_aligned_skills_50_0021/env_builder.py
975
- 933a5f52e2f3f5ee90c596c292ac5d3353e29a7682b604ab7bb5b377acd3b75b tasks/data_persona_aligned_skills_50_0021/verify_workplace.py
976
  87910d80f20fd04a2ef55901b74a011838ecaa511b3bc9366d9e5d1638c09df7 tasks/data_persona_aligned_skills_50_0021.yaml
977
  06bd01da0800305da5427aa4d1c130ec5b0760ca8ff82f62ede66efa45069b1f tasks/data_persona_aligned_skills_50_0022/_env_builder_impl.py
978
  0e56b3c683effb4992409bc0056e7936c97ec27d7719d414b08175883ab028dd tasks/data_persona_aligned_skills_50_0022/env_builder.py
979
- a41f64c73b779c1718cdc013fdbababd48d715a07a6cb1f811e9b5aae883d338 tasks/data_persona_aligned_skills_50_0022/verify_workplace.py
980
  520535d3f238adc3005c90702a47300ad23b24fb84a2e411ccd2d6340115c6e1 tasks/data_persona_aligned_skills_50_0022.yaml
981
  e31d518422ef198098f962ba3ce48bf8249f418a3a32075093fa5b2d75803bed tasks/data_persona_aligned_skills_50_0023/_env_builder_impl.py
982
  563dcc2509374b1a44788adfd8d17be421c2a8b77e5b2d6a2a82d4cbf2e29ad2 tasks/data_persona_aligned_skills_50_0023/env_builder.py
@@ -984,7 +985,7 @@ e31d518422ef198098f962ba3ce48bf8249f418a3a32075093fa5b2d75803bed tasks/data_per
984
  4bad7c53510de37ff6813ce69fbc12d290c1ad6a3cd4cbfd1d6de0fe3244c626 tasks/data_persona_aligned_skills_50_0023.yaml
985
  d35963fd45280725c1571f25d1d08fc82cfe2c7078b7172f235f73afd31d8d02 tasks/data_persona_aligned_skills_50_0024/_env_builder_impl.py
986
  8799c3bdfa812fd2986a50ebda72389e9a5aeca523e9f946cc0557afa3d0b36c tasks/data_persona_aligned_skills_50_0024/env_builder.py
987
- 21cc9c6b17a14453efc90a4847a0d6b398aa9af500c90323a47a077da379d719 tasks/data_persona_aligned_skills_50_0024/verify_workplace.py
988
  052e79a68e27e08178852a47d678a4da24baa9bd3434c20c5247c282bbdd9f1a tasks/data_persona_aligned_skills_50_0024.yaml
989
  1efafc6fb51758486d20100fd773c07a991507c921dbb94b0e452be8119faf60 tasks/data_persona_aligned_skills_50_0025/_env_builder_impl.py
990
  229064e628c13e26467710700e915658c662c7b346553272384169f31ce90f7e tasks/data_persona_aligned_skills_50_0025/env_builder.py
@@ -992,7 +993,7 @@ da9aec0b9f02b95f4b4a5e760f62297d9cd18d74efbe40ea33efaf240991c7f0 tasks/data_per
992
  7737c2b596b3d610b399a9f33a32e46fc4fecf89204ae3e2a64545a764f5afaf tasks/data_persona_aligned_skills_50_0025.yaml
993
  3c1bec3f4470b12e5d77dbd90a24ab2a76154e8f9a7830774549252db0ac4cd3 tasks/data_persona_aligned_skills_50_0026/_env_builder_impl.py
994
  933416369c3bb58ba3314503c887c6c8f577e6b737f283a4f36ac119913911cb tasks/data_persona_aligned_skills_50_0026/env_builder.py
995
- 8ae56914470cc69765be22fbd88b52d6c5d3543f26c4811d90685d18ee26108e tasks/data_persona_aligned_skills_50_0026/verify_workplace.py
996
  9c8cdb25f86d02bc56ce53359ecad12fb088425ae74ae4f0cdc4a728d0342457 tasks/data_persona_aligned_skills_50_0026.yaml
997
  53176ff4bb81e872166aa77535ea05416588fc365381b1b6fe18921468227d13 tasks/data_persona_aligned_skills_50_0027/_env_builder_impl.py
998
  ffa0c07465ab60eadad5fe2bae3013d5df6ea4e0291e72df7ebcef4ce378a0ea tasks/data_persona_aligned_skills_50_0027/env_builder.py
@@ -1004,7 +1005,7 @@ dd0b12f2830fc643f4f1fb0c4ce5942d9b09ac03645f5d74f933e8fab5c2bf55 tasks/data_per
1004
  b512a128cdce54330a0f593481fa9c49aeee361ad5fb2c752ff030f187fe3543 tasks/data_persona_aligned_skills_50_0028.yaml
1005
  ef11a48e0bff07f41db212d1e056641e6fb84bf7092d4de9f1cb12c89aac963e tasks/data_persona_aligned_skills_50_0029/_env_builder_impl.py
1006
  9147038f21e5c68c2f58069a4a80d1f69400ebe6bc7b5c745e122d32461073aa tasks/data_persona_aligned_skills_50_0029/env_builder.py
1007
- e11c99d8f991b3b9daa419e47d6379d3af911f321b3dc1173d4b99862f6daa5a tasks/data_persona_aligned_skills_50_0029/verify_workplace.py
1008
  ac86eb2860d2e0d9748d160e6b65a6b63c9df1013b9b983d54f81fdfe859599d tasks/data_persona_aligned_skills_50_0029.yaml
1009
  01b4a62c7aa5471bd101e9cff336caab0e205526f20e85068e33281b15caa14f tasks/data_persona_aligned_skills_50_0030/_env_builder_impl.py
1010
  ac362980bbf9cfb0af61bdb5ee9e393efd270650ff9c3754bcb1e5c16390b00e tasks/data_persona_aligned_skills_50_0030/env_builder.py
@@ -1020,23 +1021,23 @@ ea3faae6a1d7e79a2b97cca4b6c3f6803aa99f3f12fec72c97c7370b80df28f1 tasks/data_per
1020
  c335a6527dd41dc3305842d835c57e1adae5adb7b27c99ab6ca82927d458aad4 tasks/data_persona_aligned_skills_50_0032.yaml
1021
  ae6beef9ce164ded168785e5673e9d5ef52d2a58873256107e721f0df07d2d05 tasks/data_persona_aligned_skills_50_0033/_env_builder_impl.py
1022
  cf492a14d6ba8cb1389d6553ee3922d1d43d8e632a80f5bba1c49c8cd8841385 tasks/data_persona_aligned_skills_50_0033/env_builder.py
1023
- 521bfcaebb371d477bca2cba883f2d7527ed8e8dbd1d3f8dfb5fa04c86eee3dc tasks/data_persona_aligned_skills_50_0033/verify_workplace.py
1024
  feabc1f80335bc3347ec9c98941e726c6d0d9c80c1687471cd078c39679a2b7c tasks/data_persona_aligned_skills_50_0033.yaml
1025
  c9894b735720a06bb0437dd724047e679ab9713cf7789cdbdfea17beb19a30b7 tasks/data_persona_aligned_skills_50_0034/_env_builder_impl.py
1026
  ee0aa6e61f651e79bb533bc0cd32a830f01efd9c5aada1cbd35ccdc89f74d404 tasks/data_persona_aligned_skills_50_0034/env_builder.py
1027
- 0d200a80fa6a64d3ae9151388f3595103e43a31e800445bf04eb7ce32fc541b9 tasks/data_persona_aligned_skills_50_0034/verify_workplace.py
1028
  0427163956b3de9af9a5424289a4cce349e5538996f1f01b02bb1679bec12860 tasks/data_persona_aligned_skills_50_0034.yaml
1029
  9bb3d0e5a3664de1334501d49ea7259f3cae0acf77bb3214f4d765c4023b96a3 tasks/data_persona_aligned_skills_50_0035/_env_builder_impl.py
1030
  85c680d11531ff47dab821ace681e14aaa7f07074c471116af8225a0064a6638 tasks/data_persona_aligned_skills_50_0035/env_builder.py
1031
- b7a1ccb588a8a997c4b3821b6f8c1aa34b3e166507e53294c3bc197684f0f695 tasks/data_persona_aligned_skills_50_0035/verify_workplace.py
1032
  19d53076922ece67067be7cb16488f31d02a81e3e50ba3fa981d18872ae92f71 tasks/data_persona_aligned_skills_50_0035.yaml
1033
  0bdad4710b0ac257843288c516f7403c65a06f28981d5faa9e0fda4248e0b014 tasks/data_persona_aligned_skills_50_0036/_env_builder_impl.py
1034
  5c00d5a4240ae379b108e153f893be0dc3bc4b07d020fa1f4bdd5d20aaede295 tasks/data_persona_aligned_skills_50_0036/env_builder.py
1035
- 6f389d1ce610ee2b4f426647ec7000441edff1d8fc40952a994ce88b4fabaf37 tasks/data_persona_aligned_skills_50_0036/verify_workplace.py
1036
  d225a42bfe56bf3c4a3d943ba4fb8e04aa3c2478adb18bb40fe52247a4b0c622 tasks/data_persona_aligned_skills_50_0036.yaml
1037
  22aad0a3550173e6745e1cd169614aeaa7e508ea2ceb7c69fc1f6ccd332844d3 tasks/data_persona_aligned_skills_50_0037/_env_builder_impl.py
1038
  feb57314ff7674cbb29a97127d8364fd72a2b9b1a6843f95e6365b9d8b532ecd tasks/data_persona_aligned_skills_50_0037/env_builder.py
1039
- a9068ff071983d3dd611b2d5a2acc8617176707412c32a9815caa022b35fb4cc tasks/data_persona_aligned_skills_50_0037/verify_workplace.py
1040
  7c9a74cdcc43bcb4d4f75e761a0dc3bb25604680f69533dd63bb4c00271665a2 tasks/data_persona_aligned_skills_50_0037.yaml
1041
  01001db221ce59adac2d4ba0c2b97efcd0b0c2488964009f833ae938936fc90f tasks/data_persona_aligned_skills_50_0038/_env_builder_impl.py
1042
  45d60213b88dd3693717dd38b83fb533ed6988bb7c24c14815cbfb9ee7543466 tasks/data_persona_aligned_skills_50_0038/env_builder.py
@@ -1056,7 +1057,7 @@ a0f6aa5f11ba798790b601aa8c49c0a76f4c8ca7e087a7f237aab1a9a7b21e1a tasks/data_per
1056
  412d23bc38ad9f7aead3bb5f17d2c1637da3fa4193874c72afeb71c62350f67f tasks/data_persona_aligned_skills_50_0041.yaml
1057
  140c27a7bb83249c9e2298167294af52b8985628e87b31823c767d96f07c26ea tasks/data_persona_aligned_skills_50_0042/_env_builder_impl.py
1058
  ffbb8d53084d7fbc08206f9f75780c81246f5a4e014502fb4c067777068b1a8d tasks/data_persona_aligned_skills_50_0042/env_builder.py
1059
- c01bf548db82441bdaf99efc43738cf0bdab4f382e7b593915156e3919799be5 tasks/data_persona_aligned_skills_50_0042/verify_workplace.py
1060
  d230f00e94bc52eca65dd51512f96c1d6efeae851bb868c9c4d148edfc110190 tasks/data_persona_aligned_skills_50_0042.yaml
1061
  c075e9406486d573d4632833c728b4f60d218c14563c163faa6958c8095906a4 tasks/data_persona_aligned_skills_50_0043/_env_builder_impl.py
1062
  cbf0a4977e35a0e14fc9312d26992a6c2206ffed2ec7e4d7738b2495f53d0a6d tasks/data_persona_aligned_skills_50_0043/env_builder.py
@@ -1064,19 +1065,19 @@ a936111f462a68a3be70135a9eb1f53eef7fb6405c6849c66bb4b49a4af8b780 tasks/data_per
1064
  773bbd1b442826108f6ba98877d759011a295d8353b0b84fb6897dbac748b6f3 tasks/data_persona_aligned_skills_50_0043.yaml
1065
  55b89991abffc6b0fd53b2660764cf9724aef3c90e75a00689fbe63c926acb81 tasks/data_persona_aligned_skills_50_0044/_env_builder_impl.py
1066
  6a01c6cfbf804d535758152d64ec149c68447a159a97c13ac4ac4ae0ef794e6f tasks/data_persona_aligned_skills_50_0044/env_builder.py
1067
- ff0ceae8aac73c2d67883c01d112de819f8a401f4ff7d17ac7e5d2a6ea688d27 tasks/data_persona_aligned_skills_50_0044/verify_workplace.py
1068
  72264c7d2c9337565af2dfdf6dd496c153e264715df0ac16724f0aef9b5fce37 tasks/data_persona_aligned_skills_50_0044.yaml
1069
  962cc304260ff512b2f2600fae9ca9e139fe6b738f78d5d10ac4faeb79e0d23e tasks/data_persona_aligned_skills_50_0045/_env_builder_impl.py
1070
  883b570204c8cf928af8390ec30822fda289c49b03eb0c9c38672bad4627ba6b tasks/data_persona_aligned_skills_50_0045/env_builder.py
1071
- 070516b9309bf8ac0ba56bb7816c92b82b6551d131902b1c5b23c58709b1fa24 tasks/data_persona_aligned_skills_50_0045/verify_workplace.py
1072
  87831e1000803c11f8aa843c009acb9111b4719565e07f191b35e3a868847dd3 tasks/data_persona_aligned_skills_50_0045.yaml
1073
  ca397f3f94e7132c5224262ee863cbe9f04dc5b3c82661e2603dd0c1f8058413 tasks/data_persona_aligned_skills_50_0046/_env_builder_impl.py
1074
  c6ccdc8c10e5aea49d7b5c2c2f924121eeb9f11a1953e9b4cf7bcd272128412b tasks/data_persona_aligned_skills_50_0046/env_builder.py
1075
- 375292151b118c6d6760982a7a5ef9b416bab56ac9572b6109bf9a21a0415ea1 tasks/data_persona_aligned_skills_50_0046/verify_workplace.py
1076
  65e50bdafa9d8adf5ccd35aaf1570b7b21d3bc5fa81726c3e5090b349486e457 tasks/data_persona_aligned_skills_50_0046.yaml
1077
  d00ded6ebc6b3b6d62e6db06af2909d201e355b37cd6f9b730370a9d6cee2419 tasks/data_persona_aligned_skills_50_0047/_env_builder_impl.py
1078
  3ba4f6142e9e6719e30e69d4dd39b5335ef5d1a719af6bbf1bd989f5f7f0b7da tasks/data_persona_aligned_skills_50_0047/env_builder.py
1079
- 5e7889cc67c56eb055dbb79c3cd259c2d2ebdcf22c08d5eef48f104d944528f7 tasks/data_persona_aligned_skills_50_0047/verify_workplace.py
1080
  b22b0d5e24c65817e6c535d395cf20e9641b0b43f78e93a248f3a748cc0a40b8 tasks/data_persona_aligned_skills_50_0047.yaml
1081
  bd3211d73e553fb3a9d1c3ffad81baf1b715f6259a8e3d9bd742af362b763fa5 tasks/data_persona_aligned_skills_50_0048/_env_builder_impl.py
1082
  a0d75c1bec36c11ab36a64aabef358d67b5f5b37588e2761e330647ea54c1374 tasks/data_persona_aligned_skills_50_0048/env_builder.py
@@ -1088,7 +1089,7 @@ ed512e153c6654337d55a183c02b8fbfd7965fc5e3e49a0af89fc0673e020d24 tasks/data_per
1088
  f078c6121ad45719dd02c2e90b1b7abb171665bf20dfac83337a4868743bf04a tasks/data_persona_aligned_skills_50_0049.yaml
1089
  1bafc5e8b098ca733508d5dd1a1986da3c60015aa57f2318a6b80a47028a4085 tasks/data_persona_aligned_skills_50_0050/_env_builder_impl.py
1090
  30158949469822f94fd21d511b6df006c3767f9130a0425a58750bc3bb856171 tasks/data_persona_aligned_skills_50_0050/env_builder.py
1091
- e48125b97264a1181e528a8f4164194c30b8cd0f36ad0ab9b8c66ca726551205 tasks/data_persona_aligned_skills_50_0050/verify_workplace.py
1092
  bae66782f8c08b21ecba97700782f5ef09dc4b2c805d0705415ae6de41a7ca2d tasks/data_persona_aligned_skills_50_0050.yaml
1093
  9e65f31d924406cea0ae3ec57d5a174143672d11733f9480e0acf9d1ce1459fc tasks/prompts/data_persona_aligned_base_50_0001.md
1094
  0e92ca66c816d518f672fee4a64b357935c59b12f73c136a44a434d91331bae3 tasks/prompts/data_persona_aligned_base_50_0002.md
@@ -1390,7 +1391,7 @@ b675f14fb3a5026397130b64cd4e2ba4f7e76fa360e34ad85d5838b981d5ad32 tasks/prompts/
1390
  318e330f318d95cc36aab3076c495ddcab142476156d939c2038a4378a575aaa tasks/prompts/data_persona_aligned_skills_50_0048.md
1391
  93002945e85af62715331c5abbdd6509430ff1ae5ca8428f0ae1ede1f38624bd tasks/prompts/data_persona_aligned_skills_50_0049.md
1392
  539d7f3471c7a25cce4874b93d01f7a03cf162fbea98c16182fa8a838107f5f5 tasks/prompts/data_persona_aligned_skills_50_0050.md
1393
- e6d659e8f2f5b33f66cb54b50ced3065f1154fbd5ab2f5b75efb4167de655e6e verifiers/base.jsonl
1394
- 4f9d52e44f7195271bd07e7e0e0c5b141fb545046d278f1ad47be7cfbf894a5b verifiers/hard.jsonl
1395
- 98f92fb419e717d04a27e75e024a37ea8d3613331be26700ad6b792bb73188f9 verifiers/multi_turn.jsonl
1396
- 47b27782f0a114d7ff5aea1de765400b277c8ade64a2278b95ce8002dfdf1e6a verifiers/skills.jsonl
 
10
  937eaca759b3664669c8aeb3a0705f12d2fa66a1cb40674e425dbb97048064f4 eval_manifests/skills.jsonl
11
  2d0a464c28dc1aa380cf5740a7ba9aa76bcc33649e46ea32d8ce12ba6b601027 eval_manifests/skills.task_ids
12
  273c0a3d6342edf14abd7ea4f10f9c9179e7982455e8d99be39c0739689d50fa import_manifest.jsonl
13
+ 2ab137c076c7104a884af3249ed3d039df0493f8811e611bdec01f6bccb39eaa manifest.json
14
  12ae83d267551b1e73808354b802a7d0efcc1a3b76453e7b84c9964c4e294503 provenance/eval_manifests/base.jsonl
15
  72aeea0c6321b55982263dbd1cbc23ff114768b3cc21b3cfeab7ff70b7e00284 provenance/eval_manifests/base.task_ids
16
  cdfe914540244feb618a00470b455aba9622d94761352a174dda05826f79d040 provenance/eval_manifests/hard.jsonl
 
28
  cf3ca3b84a84ca57b8914d4fc3d08e15d04838687ae503baef3206c00888d9ac provenance/selection_summary.json
29
  48fdd4735d4a5bae570f6436e1cfcfe10ba5d236c6522da69ec2960b115670a2 provenance/task_manifest.csv
30
  46c5f60e5218f57b7dd190fee99eb81ca932e52f3f09b11fbf1b76861fa2ef9a provenance/validation_report.json
31
+ 216e7ba29a42c5bc74324bc1c0c536b5c2103a13f9eae7f05831598cd4e8257f provenance/verifier_materialization_manifest.jsonl
32
+ bb169e2015ddd2274fe010287da68bd2c3aeda72e4b1062adfa5445fc0b17260 provenance/verifier_repair_manifest.jsonl
33
  7b9957d6b41f006baa0fb5661a7b84520eaeaf2ca0baba12968c99e6e7039033 selection_manifest.jsonl
34
  6b74f75513fad3b4fd1cdebd1e2edc931602987fb6305045f614087f917354c3 skills/data-persona-aligned-skills-50-0001-legacy-raft-parser-skill/SKILL.md
35
  96ea388e187a635832d43c3306cb4c9988d57aed2cf144a92244875fb8c98567 skills/data-persona-aligned-skills-50-0001-legacy-raft-parser-skill/legacy_raft_parser_skill.py
 
397
  419c473e36b05d09253cc080d90514eb79e1730c08b20d5c25a28926f4e0e976 tasks/data_persona_aligned_base_50_0026.yaml
398
  309fe08d3f1f04be1366ea65740e959b651f91fe55d4bd13da4c41db1d679414 tasks/data_persona_aligned_base_50_0027/_env_builder_impl.py
399
  701e904057b4ff9c724910458106eee93f4f7fc03fe7469aeee72abdecaccdfd tasks/data_persona_aligned_base_50_0027/env_builder.py
400
+ feda892bdd97f025b23b1c84d6816bb184bd5fbf8b0254bc0ff7ca8bc7186ee9 tasks/data_persona_aligned_base_50_0027/verify_workplace.py
401
  adf7bda9cf224903ae10b3946e8b5ceddd73f5e6d8647f3b12f0fb1d1b5efd94 tasks/data_persona_aligned_base_50_0027.yaml
402
  99e50d5dac2bf083625b694a59f8a0f6788fef363189e9f628af378e2db19e49 tasks/data_persona_aligned_base_50_0028/_env_builder_impl.py
403
  30376ea0cf4aee7c2dec63117aa45844e1fa81511e864a4f65b0dee849272660 tasks/data_persona_aligned_base_50_0028/env_builder.py
404
+ e57b3162d38b894f5841e12ffd46b8a5af9e05823fea15b9f59b8397223a832c tasks/data_persona_aligned_base_50_0028/verify_workplace.py
405
  f836f191afc149f21bd1a50c7babd2eaee3c191ecd24415184456dd0209a6115 tasks/data_persona_aligned_base_50_0028.yaml
406
  c1b63bdd52c37439a150793741e470c9ef81c1963175ce5c3b76698704154e21 tasks/data_persona_aligned_base_50_0029/_env_builder_impl.py
407
  1cae3b31de8df3ead7017cb31c7cf86f9c80ff05175fa490d7c7488f1966dabd tasks/data_persona_aligned_base_50_0029/env_builder.py
 
413
  94493ce13f1f732cdb9b1ce467de39300de2c4ce52ada5c5213e2c7dd5079003 tasks/data_persona_aligned_base_50_0030.yaml
414
  2d38a884e90526b3985d2dfbefc313f7a92b9e490164dbf94fdbd7117202947d tasks/data_persona_aligned_base_50_0031/_env_builder_impl.py
415
  962233981532626c1d8ad5a71a3502d076a57e494890127cdfb84172efb11fca tasks/data_persona_aligned_base_50_0031/env_builder.py
416
+ 1f63100ec1169b9a0d4820d11e3d0ae6761ad39c45931b4a0659ed96a42fda2b tasks/data_persona_aligned_base_50_0031/verify_workplace.py
417
  a0c6f4af4e68ff04f39239543bd0e9b952818074c5c6827a8be58e2cd7b2edb3 tasks/data_persona_aligned_base_50_0031.yaml
418
  a50b6379bd53a7c06a8de7cd843d681def34cbf074a926d8a400ea5450f5aeb2 tasks/data_persona_aligned_base_50_0032/_env_builder_impl.py
419
  0d0659c5bd27644758e79b1644369f3af8f4caf6ac7b454f1e308cd87ed98a61 tasks/data_persona_aligned_base_50_0032/env_builder.py
 
441
  e163a7719e34c46c247540197dc4ccb9b4ad22dd6e18dbf676817ef018e06e00 tasks/data_persona_aligned_base_50_0037.yaml
442
  0a345ac8c0b22a643e0382c660b34d0bb34f9e9c768eafd84a374a8403ca7763 tasks/data_persona_aligned_base_50_0038/_env_builder_impl.py
443
  d307444652d33c8e30fc3d2cbfa33a7be79c5f4c5205d5aad21b6a5d660f62ac tasks/data_persona_aligned_base_50_0038/env_builder.py
444
+ 5013b1064701b66480f081b809798ccadeea1c845a804bd12c14c444add53d0d tasks/data_persona_aligned_base_50_0038/verify_workplace.py
445
  38aae608a4c8247283e4021b8cd3d6185fa1b34f30574b23c7268ffb461a20a9 tasks/data_persona_aligned_base_50_0038.yaml
446
  f19c5b30bd5c3a9b832b2d28d05b73994e7563a8581265b337a7ca1bf54d36b5 tasks/data_persona_aligned_base_50_0039/_env_builder_impl.py
447
  f9116c57a7886975283b29c060c9989352a11d923624c433b5fb34cae2ef26bf tasks/data_persona_aligned_base_50_0039/env_builder.py
448
+ 32e3c954591814ff3efb1388bc6667c4f7a48244d95f63d6dd84b53d3186b562 tasks/data_persona_aligned_base_50_0039/verify_workplace.py
449
  e4bb5d21f034ecfea41ce097f7f583ee9e21516288efc529f3122f1833fa7d98 tasks/data_persona_aligned_base_50_0039.yaml
450
  5046ee44b9f95c4bf1f48829c6fe39e9cb840fc55f2f424782c844e3c8bddbe7 tasks/data_persona_aligned_base_50_0040/_env_builder_impl.py
451
  fef2ef8050e166f29c89bd79ed479d597710754b3c31d00b7e11e4477c8dd9bb tasks/data_persona_aligned_base_50_0040/env_builder.py
 
465
  bd8ff94edccc5a0673c4420fadd0b5b2ed0f97dffe2bc84678220bf7ed914a1e tasks/data_persona_aligned_base_50_0043.yaml
466
  250b4807517b0b3bb86ac7d661dab733e27adb4928ca59e3d8e784613014d967 tasks/data_persona_aligned_base_50_0044/_env_builder_impl.py
467
  974c8fdcd0dc2873aeb9362c3ef9aec854580103db90148adb7e1002a0c4fa45 tasks/data_persona_aligned_base_50_0044/env_builder.py
468
+ e1eea20dcc81be5482abaf0cf2a91c77422bb57afbbc1e5767d93aa956b06235 tasks/data_persona_aligned_base_50_0044/verify_workplace.py
469
  91ec50ad850b6f8c5b9dcd6d9d6b697dbd4de208694545a4e4e4c1986d12107c tasks/data_persona_aligned_base_50_0044.yaml
470
  22503e360c49920cec2452ecb6ee9781275a8f7757bfb74f98d52b4621b2a8b1 tasks/data_persona_aligned_base_50_0045/_env_builder_impl.py
471
  ca3933e0a647e00517baf586b4608e42df835592e854e6ff03710c594f19e0f5 tasks/data_persona_aligned_base_50_0045/env_builder.py
 
489
  33e33f82ff35a6ad6cb1092e5e36b3f5939a8efe3f084955c9086b7508fcb579 tasks/data_persona_aligned_base_50_0049.yaml
490
  621b6ad772a34c7de0253650134ff3ffa2965558360f7d3064aecec375da447e tasks/data_persona_aligned_base_50_0050/_env_builder_impl.py
491
  db6022647edb92f6923877f378d15eb624578baef866632d74b05166a7cd908b tasks/data_persona_aligned_base_50_0050/env_builder.py
492
+ fb4cbd28d3076fac5c11fccd628c7b145826e672da85bf79d898aee28913b0cf tasks/data_persona_aligned_base_50_0050/verify_workplace.py
493
  39985f7f65d42723fad48a0f4f9485351aeb9d3fdba74d11987c403a5a2a5c30 tasks/data_persona_aligned_base_50_0050.yaml
494
  9892213853ca4c0261fd68c4cb4e247eaa0b21430d22b5209ea1cb29b1f4edfc tasks/data_persona_aligned_hard_50_0001/_env_builder_impl.py
495
  3e61d1d76042916afa0bd62010e86d59370e0d27c26971e7e9f3905d86b78886 tasks/data_persona_aligned_hard_50_0001/env_builder.py
 
501
  b14d812ae12f36674a7df36748145b7beaa7806f8e15ab0bedfe6a982e8ac1cc tasks/data_persona_aligned_hard_50_0002.yaml
502
  3663abd9569ead92bf55f91bc1afa38f9b3cb9983f1878204bf36b10bb755a1e tasks/data_persona_aligned_hard_50_0003/_env_builder_impl.py
503
  0565d7f925923ff2026cf8f268ca40a8e3cb64f63d0465121b5d9ba1d8743c8a tasks/data_persona_aligned_hard_50_0003/env_builder.py
504
+ 4b1d8d37a1e133836fb006c15aa83d9ff478dbd7419814b1edac013a355355ea tasks/data_persona_aligned_hard_50_0003/verify_workplace.py
505
  88ff6a2515b4135b23b5f0d99c444f0e7bc6752d16d17ed1c1679b6c55a40906 tasks/data_persona_aligned_hard_50_0003.yaml
506
  f4d62cc9d321d62f6a3dd231a882bfc6731bb6cc8ee67be55dbc97c0307ec002 tasks/data_persona_aligned_hard_50_0004/_env_builder_impl.py
507
  c3b2349fa5381acac97516862d5853aa61853adbe3dd0f288c5b78fa3005fcb4 tasks/data_persona_aligned_hard_50_0004/env_builder.py
 
509
  8bfdaca55308c27902185f964ebea07c47516010a236af3b7bbea5c65b394133 tasks/data_persona_aligned_hard_50_0004.yaml
510
  9377b5cddccdc96cffc589184fadc5dc44828a84a89d023a44ec86bc4478745f tasks/data_persona_aligned_hard_50_0005/_env_builder_impl.py
511
  ec61a256c79a4cf56633c6cd1e401203c96ef27ba8686fc8ecae56a2a3225a45 tasks/data_persona_aligned_hard_50_0005/env_builder.py
512
+ aff1accf0dcdbb61d911f7eef88f979baddb6181ad0508be6b29d06bfe912fe6 tasks/data_persona_aligned_hard_50_0005/verify_workplace.py
513
  ef459e1e02cb3ed81b88ec63e8e574199b4d0aead843d3627f9d634effa39157 tasks/data_persona_aligned_hard_50_0005.yaml
514
  1b8d9ed1662013c6063095a9dee57f4eb45e90e52aaccc23665e686d7d6ee285 tasks/data_persona_aligned_hard_50_0006/_env_builder_impl.py
515
  4de8a87722a1a52230c66a4f518b0f2dc71e9dcab64c0d4ce2a5928f3bf408ba tasks/data_persona_aligned_hard_50_0006/env_builder.py
 
521
  93ed4d1565625e3807ab2bd958b08f13eefdff0177fd2a3976191dac1559c5ff tasks/data_persona_aligned_hard_50_0007.yaml
522
  26e506a445f77a0753ad8925746c99bdc9504359ba99d6496a7cd39c1303f7e2 tasks/data_persona_aligned_hard_50_0008/_env_builder_impl.py
523
  342c106b617b8540cacc25ce925991e51f53b3cf7f8466705867de9aa8813c48 tasks/data_persona_aligned_hard_50_0008/env_builder.py
524
+ cd7dd3eca81159f53f018bf176e0711db8cfd37ddaa1e5741b7e6c7d7bf718b3 tasks/data_persona_aligned_hard_50_0008/verify_workplace.py
525
  9272346004462fd9b33b9ce938153b90b4c3ba63a20a4b3f4a238e0ae89b8b9b tasks/data_persona_aligned_hard_50_0008.yaml
526
  81d367d74e1ab93dc2c40117acf662ee1df3e59d97836a40e7694d919ae85258 tasks/data_persona_aligned_hard_50_0009/_env_builder_impl.py
527
  4c89bb7b35cbcede49935ea0a78d6ee5adf3225738523fbfe3d0161328ba05d1 tasks/data_persona_aligned_hard_50_0009/env_builder.py
 
529
  0e7bc5aa1ee8327bd92ac848ac58551cef3542910c1f8816761cf79dc307d456 tasks/data_persona_aligned_hard_50_0009.yaml
530
  aaa724cf61cbf07ae648089a1c790880da4969085fa5b6fc1b49b42134a23b66 tasks/data_persona_aligned_hard_50_0010/_env_builder_impl.py
531
  a0b86d0d17662a9c17cbd783917140943b55b5b4264df71528558781dfd36ce6 tasks/data_persona_aligned_hard_50_0010/env_builder.py
532
+ 2f833d4fa9d9d12cec8469aa179fb467f756b1356358ee93769a28287f90c87d tasks/data_persona_aligned_hard_50_0010/verify_workplace.py
533
  c1a09ff9086f6ae5a6150e807b971374dde5d1f90b998701c0b189fb4732b446 tasks/data_persona_aligned_hard_50_0010.yaml
534
  06d42d54408f75e93d1b8be591f6bf259037caae86064f98f335992f619d6c3c tasks/data_persona_aligned_hard_50_0011/_env_builder_impl.py
535
  316d452fdba8ec3f9f774b94f8a2ea6d898d280b133dff09efa1c6cc55e61cda tasks/data_persona_aligned_hard_50_0011/env_builder.py
 
561
  d04f43e4c15c4c935c9586b61e948637ce13668e1fc131f3ec326193df779e40 tasks/data_persona_aligned_hard_50_0017.yaml
562
  17fcd2a10401eb20c0778932a134f08aa739d51566965e4d35ba01be72a0f734 tasks/data_persona_aligned_hard_50_0018/_env_builder_impl.py
563
  98944163bb3ed53c6d7850cd088c09f15b9a634c1b03e4b0ccabb73fc7fe07c9 tasks/data_persona_aligned_hard_50_0018/env_builder.py
564
+ 59ba89a4abba5b10c2faa48c42303873a2d9a2aa84df4b56f5b865b70e9d3cf0 tasks/data_persona_aligned_hard_50_0018/verify_workplace.py
565
  5a374e07aa2bbc2ecd36b7f8706e293b1552920462256a87b4cfbc717d6e12b6 tasks/data_persona_aligned_hard_50_0018.yaml
566
  91f2a954bcc89ee384a12b36a45675d6d1af572f8c88d1ca0a534914592bff70 tasks/data_persona_aligned_hard_50_0019/_env_builder_impl.py
567
  765b8babb18dc5d8b9c7abc5cf3c8b527c8b3dc0718bc58486e177d68a812f70 tasks/data_persona_aligned_hard_50_0019/env_builder.py
 
589
  6abe823a388bb7d9b4178bfe63c1364714a326172c110b39feaab026ca64665d tasks/data_persona_aligned_hard_50_0024.yaml
590
  10e5ab67e724c6eff45567a16296e426583b4a01191c5c9657c4f053ffc9672e tasks/data_persona_aligned_hard_50_0025/_env_builder_impl.py
591
  9623bfa7c6debc23c10fd65680189bc336c74820b51edf66d13a3256a345c3f4 tasks/data_persona_aligned_hard_50_0025/env_builder.py
592
+ 50bc50580e9d36b60dba7fe006a2b5b08f6e9ff8c0aaf4b21a15095bbb01ef78 tasks/data_persona_aligned_hard_50_0025/verify_workplace.py
593
  aa6e710116826c1faec4c6db3ade51b804c0ed28f3d1786c25c6c9055ba297cd tasks/data_persona_aligned_hard_50_0025.yaml
594
  94dce0fdb078c2c0978b2d755a1ed677f12ad86f1b01eceb4a99bb2f77cbf766 tasks/data_persona_aligned_hard_50_0026/_env_builder_impl.py
595
  b5a5760af5aa6213db8ae77365d5a324260fbeb05ac38f6183bed96d2497e3a4 tasks/data_persona_aligned_hard_50_0026/env_builder.py
596
+ 40c3f225f245531984d172bb8601985d0e7ae1a37ed261f3c2ff7c9ed8061db1 tasks/data_persona_aligned_hard_50_0026/verify_workplace.py
597
  94d37e64629a0afc19595978c77be604fd0f69e8d021eecebd606f8697e37e44 tasks/data_persona_aligned_hard_50_0026.yaml
598
  6d71034b3da4d856404318ae3e8c6964039c746d31568bd001120f9f2ee323d8 tasks/data_persona_aligned_hard_50_0027/_env_builder_impl.py
599
  e916a151a5dfc1216d3126e1fdc9d3e49d4ff21dfd0d8e1a69c481b99e652844 tasks/data_persona_aligned_hard_50_0027/env_builder.py
 
605
  e9482f5c8e5e45ac6ce070706b4b16229f5411a3c59c9d266d20dcd4dd40a7bc tasks/data_persona_aligned_hard_50_0028.yaml
606
  641130cbb094d46c4f6a6453bf48c4da4223b1c6b2f11378cea254a3d1550cd2 tasks/data_persona_aligned_hard_50_0029/_env_builder_impl.py
607
  bedb04030f528ca6141a3a8d98870e6298ece7296c1e9427c6ef24f155292030 tasks/data_persona_aligned_hard_50_0029/env_builder.py
608
+ 060f20e68fe9815f2116dd6741b70e4d026934e6126a53d32fefda110320f392 tasks/data_persona_aligned_hard_50_0029/verify_workplace.py
609
  cb3cf4435c2d8f1b84430fba628f17a187d07500152b6869ef58ba690486596e tasks/data_persona_aligned_hard_50_0029.yaml
610
  57ec090bb1439b10bebbe4ba8effc0dd0251169a5c7afe6da44097428359edbc tasks/data_persona_aligned_hard_50_0030/_env_builder_impl.py
611
  4c060230b3d775a246e7917095863cff5352ecf49aa0861d6ad5e4b6468337e5 tasks/data_persona_aligned_hard_50_0030/env_builder.py
 
613
  653f70b05c6670ce5a312a40ddf98974ffdbeb079b10a62d5755dc7da3b4b47e tasks/data_persona_aligned_hard_50_0030.yaml
614
  8a7abefe1106481b16b296adad117e7c789d97297a674e8b734d8079718d5adc tasks/data_persona_aligned_hard_50_0031/_env_builder_impl.py
615
  1208d3209535d6a643c81fe45d9b2b2399e2c32b762ab32319642b12ee9d5040 tasks/data_persona_aligned_hard_50_0031/env_builder.py
616
+ 25719334b26cbd2147b565a8e9b31470d843c37b37498e9f3ff5f63b54c95b63 tasks/data_persona_aligned_hard_50_0031/verify_workplace.py
617
  1b61c0f4805f412a3286e128d8307900ac900e769e4e146df828df3f9be32932 tasks/data_persona_aligned_hard_50_0031.yaml
618
  72c58b2d69f62db6115a3b44e28dc24b6a8478a8c2a50d78e413fbe588b2338d tasks/data_persona_aligned_hard_50_0032/_env_builder_impl.py
619
  462064396cd8a259d88b6dbdc0be51924d3119170343c1871e159d51b9bab1e2 tasks/data_persona_aligned_hard_50_0032/env_builder.py
 
645
  d6f4dac47bcfa65828f04f5b912efe77dda129fc70f4772512092faf7b764db9 tasks/data_persona_aligned_hard_50_0038.yaml
646
  cee58c8a384c84e40e1dafdc1dd9bf1b8a8cb45bb405b164856e84d10b164b59 tasks/data_persona_aligned_hard_50_0039/_env_builder_impl.py
647
  736d65347297195a85d31a4601ad05f4f2b5798bdc24708655c97b8055c34ab1 tasks/data_persona_aligned_hard_50_0039/env_builder.py
648
+ b2649969cce6d75abbce892cdfa82d917a1fc2eeb398ae41b4bf4666042d1663 tasks/data_persona_aligned_hard_50_0039/verify_workplace.py
649
  03a305940543100215db9305eea51043bab58f331ff3c821170f90ceabed5f70 tasks/data_persona_aligned_hard_50_0039.yaml
650
  cccb54a1d4689bc2d39a5d82f25367675db0b86f9081f6e754b69ce8e6793885 tasks/data_persona_aligned_hard_50_0040/_env_builder_impl.py
651
  4aeaecc382f24f28cbb616749ecc37a92c699f6bf79e38a2a46c518a1f20d972 tasks/data_persona_aligned_hard_50_0040/env_builder.py
 
653
  b93dce571e0ac493062d2fad51245f215768a562db36abe2bb8e50ca2b80e3ae tasks/data_persona_aligned_hard_50_0040.yaml
654
  368d0e34bf22c88838c5d021770c32be32410d872885d1edab41fa9d627de4da tasks/data_persona_aligned_hard_50_0041/_env_builder_impl.py
655
  7cd26a3a972cd04ff8f87382186ef63e0847a35f52c8f96098480dca21e66068 tasks/data_persona_aligned_hard_50_0041/env_builder.py
656
+ d13838560f93ebd778833970e1379741ae9440250192886f95270526dd7ff84b tasks/data_persona_aligned_hard_50_0041/verify_workplace.py
657
  1eacd7393d74e2a684d26214b52dc96e71f2d5ec202738e725b33c362f42c048 tasks/data_persona_aligned_hard_50_0041.yaml
658
  fe55b1d0bfd298402f6a6dc4f979acff5fe3510cde4743ad3a7884f7fdf90bee tasks/data_persona_aligned_hard_50_0042/_env_builder_impl.py
659
  5fd181c6b52c2b26963900fcfb2a6636664a56e8adce3e8bf70797cc0b119d47 tasks/data_persona_aligned_hard_50_0042/env_builder.py
660
+ d107aaed634d832358474a7acb298a28f57e7255c7c8df774ed10d2316d6c5c5 tasks/data_persona_aligned_hard_50_0042/verify_workplace.py
661
  c61d5a1a8a457e950266bbe86a5dc1e601baa0c0f63eed573679581b0f6a3ed1 tasks/data_persona_aligned_hard_50_0042.yaml
662
  ddd0c64952ad98b7d527bab20b85db3c2f403102ae5fb1e687f2aeab46dfd24b tasks/data_persona_aligned_hard_50_0043/_env_builder_impl.py
663
  74c8965b40c3a1d0903a6084542ba7e77fecbc8c4a9cd0b1a91cbe65dcfb7b2a tasks/data_persona_aligned_hard_50_0043/env_builder.py
 
665
  9acfdb19b5742bd89764f244a16d4285f6c534c363cf619743aa7cc77cd2243d tasks/data_persona_aligned_hard_50_0043.yaml
666
  cba1af8dcbb9be4e8ce767e2bf32bd4712022eac1a08d45cec3a77cdfcce5db1 tasks/data_persona_aligned_hard_50_0044/_env_builder_impl.py
667
  9f2e91e8a8c10182bbad2884c67327bc1d05b0bfcd774167c2b9eb30839dc30c tasks/data_persona_aligned_hard_50_0044/env_builder.py
668
+ f023edca3e1f5fb0e827232255b3bd0b69cbf2a980d58732d911eb1cd3427e76 tasks/data_persona_aligned_hard_50_0044/verify_workplace.py
669
  ce6178011783f0c1376190f58110c5d5243259bde9b0335b85b14423b32fe60e tasks/data_persona_aligned_hard_50_0044.yaml
670
  2f7c3b1ba5149edfa9f168c3ad807326df9f8edc800c9b39e2a7d8c0cf61ae67 tasks/data_persona_aligned_hard_50_0045/_env_builder_impl.py
671
  d182c408e7279f400601c244e2f1c7a71801ee692e981dd0e377321386e4bd49 tasks/data_persona_aligned_hard_50_0045/env_builder.py
672
+ 7d5436457c8aa7af655deb5dcdb6d6f9afadda7b91f7a61c63a5b2c3098b2428 tasks/data_persona_aligned_hard_50_0045/verify_workplace.py
673
  355d9686d9318ca38b287079a45d4a50a714b19e744ec1bcc49517f35c07af68 tasks/data_persona_aligned_hard_50_0045.yaml
674
  d31aefed27596025b838d8173ee37ea7878873a4503fe42014cccbd2f60fe200 tasks/data_persona_aligned_hard_50_0046/_env_builder_impl.py
675
  9843e2b501ad3be64b7e6489b13804ff5bf44de0755d4780540bde37516e96dc tasks/data_persona_aligned_hard_50_0046/env_builder.py
 
685
  00c24ba621546cb257268342831b204cabb8062991524828b527fcbc886abf62 tasks/data_persona_aligned_hard_50_0048.yaml
686
  bb6cc3f6e6c6aa37a8f7befe92c1610110b73cee53fc880c83bf14cae5f787b3 tasks/data_persona_aligned_hard_50_0049/_env_builder_impl.py
687
  2d3389cd12b5cac19e2870170bfc7cd0262f8cd7d3992f9fe206933b822767c9 tasks/data_persona_aligned_hard_50_0049/env_builder.py
688
+ 3d13966487bf7a2eb8e03212f371283b53bb878055b76f7181f67e9d8991ab47 tasks/data_persona_aligned_hard_50_0049/verify_workplace.py
689
  8c6f40192de673eda70c90d6d49289ab51f85284861eb2a90063c50713ba26eb tasks/data_persona_aligned_hard_50_0049.yaml
690
  d131ba8ad15605f073159e5cfa8f59d1f61441775e05ed0b0d3fbaddfe22e956 tasks/data_persona_aligned_hard_50_0050/_env_builder_impl.py
691
  962bc6d415f84f358a93f98c58566b35d4583c848105e62385653afc8ca4a050 tasks/data_persona_aligned_hard_50_0050/env_builder.py
692
+ 5d9db50eb3750bef577de2d58f7c4ce5562993ee5fc7b2fcba0ccb8622a4b27e tasks/data_persona_aligned_hard_50_0050/verify_workplace.py
693
  f246d3283b5505e1dc14898df39dd637dc3f463149c433f47df715f66e462b92 tasks/data_persona_aligned_hard_50_0050.yaml
694
  e0870061523156dcf50b3d061b41eb0b55f4f1cd500110c6baa21cc8b6c7d62f tasks/data_persona_aligned_multi_turn_50_0001/_env_builder_impl.py
695
  aa5693f630e9cb100b535d7fa4a218b909ab9ae365224b51a708dc2cbf566fff tasks/data_persona_aligned_multi_turn_50_0001/env_builder.py
696
+ 3b646ac7afb34a7dec3d8cc0e1d65b9703add5f6965d39b653bcd42430ecd8e8 tasks/data_persona_aligned_multi_turn_50_0001/verify_workplace.py
697
  333636fb22be35889c8dd4548a374dc11246e3a3fc0d9a3267fbf49fb2bb5966 tasks/data_persona_aligned_multi_turn_50_0001.yaml
698
  94846961de34d7ac9d141af76fde4d100e8d7fc65a079d00104365d76bf086ab tasks/data_persona_aligned_multi_turn_50_0002/_env_builder_impl.py
699
  a7ce68a93347109f983ed0429b28ab540ea0517d3c20dc44e29a143fca98cac3 tasks/data_persona_aligned_multi_turn_50_0002/env_builder.py
700
+ 4e6460e60560e937d0542da9b5e5fc7e1b9b759ff2da417ae0025a5217933c60 tasks/data_persona_aligned_multi_turn_50_0002/verify_workplace.py
701
  e81caf1948c591f769bb11aa1910b78f8d01f2dc9c886dc1673787f01dcb1f95 tasks/data_persona_aligned_multi_turn_50_0002.yaml
702
  56b62385dc8409dc5fa0a18033446b9ecc250f93dba955aa803e0bf174f499cd tasks/data_persona_aligned_multi_turn_50_0003/_env_builder_impl.py
703
  005b608483709080698b9054c47786e7efe90831cf88c98caf750fdfb8a9006c tasks/data_persona_aligned_multi_turn_50_0003/env_builder.py
704
+ 357054ff61bdbbab0b787046f05dc4d9689c7900732d54f6b4c1a84d8d7cc2df tasks/data_persona_aligned_multi_turn_50_0003/verify_workplace.py
705
  b27b9f0128955399af5ac73b2ffbc926839ecdca9cec6be21d9a93f4f3b7892e tasks/data_persona_aligned_multi_turn_50_0003.yaml
706
  8e39e4e7954d29970b56ec7099c3d60fb38884945628454327845c041301d57d tasks/data_persona_aligned_multi_turn_50_0004/_env_builder_impl.py
707
  daa0e5221bcb2406668d13596243e86754b65490b3ac430bc5fe0e02b2772d4b tasks/data_persona_aligned_multi_turn_50_0004/env_builder.py
708
+ 1ed2b2584e902e7fdf6c7679aa49d6971edd430f134463ba7602e31b801c803a tasks/data_persona_aligned_multi_turn_50_0004/verify_workplace.py
709
  9c40b0609198ddabb5faf5c1438dadc8c8338e5866783b968c3c6f851ee50c94 tasks/data_persona_aligned_multi_turn_50_0004.yaml
710
  cba2b5f2d524c3ac229b241b5ecaadd9bd4851ee085061bad847f6f35b103198 tasks/data_persona_aligned_multi_turn_50_0005/_env_builder_impl.py
711
  d572d99fa9e6914cfa62bb790a95c896734f1582e65b7bca58cba7fda6d678e1 tasks/data_persona_aligned_multi_turn_50_0005/env_builder.py
712
+ 0dd5d3bb9048ce47ed9f6f49f3f0a303d5d7e9cc5b7852e3e9f258ca544c3e95 tasks/data_persona_aligned_multi_turn_50_0005/verify_workplace.py
713
  3a8b7a38d884dd884117035c4c1c0c1170ea9a22aa7635f0518bc5eea72d9bd0 tasks/data_persona_aligned_multi_turn_50_0005.yaml
714
  1dc41fe560925c3bfb6b1f833364d2b6c26714b35144d33c27ed692e2d1ee9a9 tasks/data_persona_aligned_multi_turn_50_0006/_env_builder_impl.py
715
  18bbd5412ba2f4c574161cb6c570ff8566d8d1376acdeab1fb4c0dcf4705e423 tasks/data_persona_aligned_multi_turn_50_0006/env_builder.py
716
+ 00f7360aee7652ae754bd02912306fad7399781cd8f81fa7c48ea5864985543a tasks/data_persona_aligned_multi_turn_50_0006/verify_workplace.py
717
  c5b7cde4c2402bf0c72663da6d8f792178cc729f3a2da8383202a01ef07ddb78 tasks/data_persona_aligned_multi_turn_50_0006.yaml
718
  53da96cb334974bb68f6379f6b5ae23ce42c9ca5ac8c11e0e167f9122944346b tasks/data_persona_aligned_multi_turn_50_0007/_env_builder_impl.py
719
  6a93ba55bdd0b829084ca629904551c0f79a91aa457a6042bae3d225b5f48cc0 tasks/data_persona_aligned_multi_turn_50_0007/env_builder.py
720
+ d1f92de1a92a6ddfdf294f32e1b14e6f209e217351d1e7c376493072c6c9fe76 tasks/data_persona_aligned_multi_turn_50_0007/verify_workplace.py
721
  2cbe9b2c88264813ea9bca905cbcf16ea7a3ca7a18a00c70ff422ec4e85cf9e2 tasks/data_persona_aligned_multi_turn_50_0007.yaml
722
  9bee8ee455207c056c55f34ead641e97a7c83d0d91e5f5ed582eb4e27bf1ba9a tasks/data_persona_aligned_multi_turn_50_0008/_env_builder_impl.py
723
  8b594c39ed78a5714a289f4478075ca59bca28dd66e8d5cf720b2e7ea2b0390c tasks/data_persona_aligned_multi_turn_50_0008/env_builder.py
724
+ 58d93c993afac3a9bfaba8de8459153a4d74b02e3270ce4e862df49f85765ed6 tasks/data_persona_aligned_multi_turn_50_0008/verify_workplace.py
725
  7f4e82abd2ce0b7e78134d30112e67332d32a7f34a37f882a94f39a6f7a832a3 tasks/data_persona_aligned_multi_turn_50_0008.yaml
726
  19eb062b37e50c552972258444289b269b371cca4b707fb95d98345a27c41da5 tasks/data_persona_aligned_multi_turn_50_0009/_env_builder_impl.py
727
  3fd3bd1560b093b7795bffac6256e1378d734b34692bb6fb4155566a3fd202f7 tasks/data_persona_aligned_multi_turn_50_0009/env_builder.py
728
+ 70668ad96844187c7b062535439cea3b28a9312a989929dd6fc7df117e3a94b3 tasks/data_persona_aligned_multi_turn_50_0009/verify_workplace.py
729
  fb0862f31973d092ce9bdd17a3e44ba234b2f1cf69b25550ba600a46df727b96 tasks/data_persona_aligned_multi_turn_50_0009.yaml
730
  bba2fb8f620eaeaa151bc313d9e61ad95411dcf4ea11fa837626dc0b6fc1e0b0 tasks/data_persona_aligned_multi_turn_50_0010/_env_builder_impl.py
731
  56b05bc689641318b690c0425450f367c608b35841f7f5b98da729b75d73cc4d tasks/data_persona_aligned_multi_turn_50_0010/env_builder.py
732
+ 2f9a7370864cf7e3c7c8f9afa5a5b4a56561bf8daced7882f741c1779353a503 tasks/data_persona_aligned_multi_turn_50_0010/verify_workplace.py
733
  b3ecbd7315c94c81aae322dbeaad3cb8e43a8780c3eea940fa7aeab7b1f14878 tasks/data_persona_aligned_multi_turn_50_0010.yaml
734
  8003c68cb6d637cc88ca863aed65fb344e817be2360e6324b432c73101fa5c40 tasks/data_persona_aligned_multi_turn_50_0011/_env_builder_impl.py
735
  b4625423f6ce3868ae90cd07bf017dfa3d85578f2926006180e03be729d41ab6 tasks/data_persona_aligned_multi_turn_50_0011/env_builder.py
736
+ 0b4de3417542b68dd16e5b2625558f4ab69fa45d4272f5ee59bf3e6d68bb9212 tasks/data_persona_aligned_multi_turn_50_0011/verify_workplace.py
737
  3a09c0e7a11d6474f6072bc31c2982655c55896bee8499733c8f32f441544719 tasks/data_persona_aligned_multi_turn_50_0011.yaml
738
  e4b00c3dc279210b49d7ff3eebde96ed00f689d4cbbe0f9f1db8f8967ef69549 tasks/data_persona_aligned_multi_turn_50_0012/_env_builder_impl.py
739
  0ed06754dc947f478769feedf9ffc69b231b19bb3930e80913cc92952307cc87 tasks/data_persona_aligned_multi_turn_50_0012/env_builder.py
740
+ 66857a2e701c7a967dfeac3fa35996ad74681db8a2ccc906af79a3d35424ae8e tasks/data_persona_aligned_multi_turn_50_0012/verify_workplace.py
741
  682f72ce4ba4ac87b0b7f202d6dc4f2f5c855843264299130cbfbb8ab8ccc651 tasks/data_persona_aligned_multi_turn_50_0012.yaml
742
  220477253bce809ca1035473da392466a5009633b9312964979f664c7318861d tasks/data_persona_aligned_multi_turn_50_0013/_env_builder_impl.py
743
  36ee6290bea8542c8dacea5d70d9ae0ae967b353466240282629cc9dad6e11ce tasks/data_persona_aligned_multi_turn_50_0013/env_builder.py
744
+ 9dd090bf0a85dfbb78e2f986d81c97f0222cec71d69c8a929d532553fa9e5770 tasks/data_persona_aligned_multi_turn_50_0013/verify_workplace.py
745
  87163756033db05d116adbaf83171f3ff77e02c9cd8a0715f0587f68a7b91409 tasks/data_persona_aligned_multi_turn_50_0013.yaml
746
  60e5cd1d6b6fbcb2c01eafc5fe8bb0c59c74d9d1e32105a207191fe713f6af4a tasks/data_persona_aligned_multi_turn_50_0014/_env_builder_impl.py
747
  c42207780a8226358470b5328b1ecdd062aa2cd5877bad0de7cfa6dfcf7b44d0 tasks/data_persona_aligned_multi_turn_50_0014/env_builder.py
748
+ 47334ce5fc53b43135758cb5844945efacae5236c52596c2749403908a7d497c tasks/data_persona_aligned_multi_turn_50_0014/verify_workplace.py
749
  3849aca1cef3ca1d7635b8dbd325d970d1ec6207d45afeffc7ca619e4bddc0a8 tasks/data_persona_aligned_multi_turn_50_0014.yaml
750
  7c6577faff4bebc7961e19db2f249c1916f909a308f89922eea394380206e1f9 tasks/data_persona_aligned_multi_turn_50_0015/_env_builder_impl.py
751
  f7a54700886af3193ddf9859affea8860dfa76c1c39182c3a02d283d50dd18d5 tasks/data_persona_aligned_multi_turn_50_0015/env_builder.py
752
+ 0fd9b564c3e79ab33aa2e7196c32ea4fbf836059ed3fabbd7bfb9cbfbf98d986 tasks/data_persona_aligned_multi_turn_50_0015/verify_workplace.py
753
  a4ca347eb99244e243ca85bf6e4206760df3028349725e09cd91ca03cbed8ecf tasks/data_persona_aligned_multi_turn_50_0015.yaml
754
  09bca3252d8d83eb9f5eef55eb377a6410dd4d8867bf99cd7f067ee4d12ecaa3 tasks/data_persona_aligned_multi_turn_50_0016/_env_builder_impl.py
755
  574ef7bae7593f94697b980b097dfce54aca011d6dd12f4ec33d13833c975467 tasks/data_persona_aligned_multi_turn_50_0016/env_builder.py
756
+ d5ebc81a5dfb6c368b64db9b0b7b7043e8f71edd46e77d857337e11a69055f74 tasks/data_persona_aligned_multi_turn_50_0016/verify_workplace.py
757
  47f0f2a12c50e596768ba756aaef4ecc35c8a292f023c4e2b5afcdb36275551a tasks/data_persona_aligned_multi_turn_50_0016.yaml
758
  c643e2ad89f0ec3bec6e0dd36b61da9b8b7f87b3a6b09913328ab0783da9fcda tasks/data_persona_aligned_multi_turn_50_0017/_env_builder_impl.py
759
  1d98fb6b956e9fcc57cd8ac436ccb8bb72164fa6e81c68dbc7ab7e5c20e7ba59 tasks/data_persona_aligned_multi_turn_50_0017/env_builder.py
760
+ 97b32d10e889a9d97cca9c9c071e8f4895587e834d4509bfd61fb0690ffa0879 tasks/data_persona_aligned_multi_turn_50_0017/verify_workplace.py
761
  47e9b166b2b5f65d19bf778415e950e06d479ad8500b2c362870ad086bdc9ddd tasks/data_persona_aligned_multi_turn_50_0017.yaml
762
  2c734b483f1be0469fda215bed6657704fac8cd0949a58107d4f49b4f7fc5676 tasks/data_persona_aligned_multi_turn_50_0018/_env_builder_impl.py
763
  4797fb5c7d9af2c55afb658e3df552ca12d3489305802f658da44e2a8300045c tasks/data_persona_aligned_multi_turn_50_0018/env_builder.py
764
+ 5c436f3b2290ceb0a16b7d5a6c860e086facd5910f9e021709b27a22b5e70bc1 tasks/data_persona_aligned_multi_turn_50_0018/verify_workplace.py
765
  fb8116b17ce62d333453921784668a580f7f72e1dc8153ba39d9254bb48fc8bb tasks/data_persona_aligned_multi_turn_50_0018.yaml
766
  5c619a88b5de3cff6e4d453ca9bc26c80ab888a9074007c10fbd4e9b8e3769e2 tasks/data_persona_aligned_multi_turn_50_0019/_env_builder_impl.py
767
  a79017e9c78017b90ec33b2d8a7d56d9f5f7fd22016a4cd89755045d35c29e7f tasks/data_persona_aligned_multi_turn_50_0019/env_builder.py
768
+ 4dfa7d1a72fd7c9d17376fac8299423cb145f765db569cdcfbe2e585deb71f1e tasks/data_persona_aligned_multi_turn_50_0019/verify_workplace.py
769
  70d6295d4f3bfcad9088cbdf1e1950a1fa51d03b880a6d1b5c886c4f6819793e tasks/data_persona_aligned_multi_turn_50_0019.yaml
770
  c738370ce6a9a4e9ced171c1fe8cecf0990082c2787269e00699d4903c100cad tasks/data_persona_aligned_multi_turn_50_0020/_env_builder_impl.py
771
  8f313309971ecb7df8ce2132f41c2661137c308a82a62cd92840ba1b906f1270 tasks/data_persona_aligned_multi_turn_50_0020/env_builder.py
772
+ 8e70f51e20faea712d9d0adefa4315de1d05ef43a83bc08e2f9330b598ff2318 tasks/data_persona_aligned_multi_turn_50_0020/verify_workplace.py
773
  85c811f41301ba28aea1c24422ef98b822d4b22a9d564bca89a6b88bcfbab2c1 tasks/data_persona_aligned_multi_turn_50_0020.yaml
774
  3be98a23963dbc1e303348567aee69aea0aee4fe028bc96f2599d5e7529f2fea tasks/data_persona_aligned_multi_turn_50_0021/_env_builder_impl.py
775
  0edfd1c1ac717a0729da0d981c82c4ffddfbc0184921e75b2fb7b16da8ea2b62 tasks/data_persona_aligned_multi_turn_50_0021/env_builder.py
776
+ 4e61d7af7a2a92ed6ee5c10b3d9c085599c63fd3695f1577314911d25a28c7a0 tasks/data_persona_aligned_multi_turn_50_0021/verify_workplace.py
777
  44754ccd1710c477dfca7ecea24b3f8e2dd37129598b6f855335929db6828915 tasks/data_persona_aligned_multi_turn_50_0021.yaml
778
  76314efd0543d714564316e3f39d8590f93b1644a3a2c33d57287c21f9596b37 tasks/data_persona_aligned_multi_turn_50_0022/_env_builder_impl.py
779
  703067eb33128d48bda186519dfb9166f7713746138162b0b34ad7c3a8483370 tasks/data_persona_aligned_multi_turn_50_0022/env_builder.py
780
+ 4b2c59c6cd7a74c8fc2194fa90cc612b7be86aa8d4aa9d4b326dc05526c531f8 tasks/data_persona_aligned_multi_turn_50_0022/verify_workplace.py
781
  2856c77f67701b8a42c05f22160dde6701ca2be1ec31af7494056431c4c90c34 tasks/data_persona_aligned_multi_turn_50_0022.yaml
782
  d7a1f604c95694bdfc8f349a31b774f006747711a6df4db92bd3c3aaec990275 tasks/data_persona_aligned_multi_turn_50_0023/_env_builder_impl.py
783
  a09ea95b569dff0baff3c85af3192fdc8d2697beb4d7d275b1286b9a22167682 tasks/data_persona_aligned_multi_turn_50_0023/env_builder.py
784
+ f04cde53e10c053fba755271514380ff0b7465db43fabde27c57be07fe4213ba tasks/data_persona_aligned_multi_turn_50_0023/verify_workplace.py
785
  65ff7137784414b5135c907b66c1325b88c0bd4349868af5d69418446fdd4ff8 tasks/data_persona_aligned_multi_turn_50_0023.yaml
786
  bdcb9cfb34d07f1f005497b3e51eb8a3ec486accbb314bfa3549a37f06fb2b56 tasks/data_persona_aligned_multi_turn_50_0024/_env_builder_impl.py
787
  fbc69703bb119564f71c801242fdd7b998ddaabf708c348c39907f8e9a152f55 tasks/data_persona_aligned_multi_turn_50_0024/env_builder.py
788
+ 76b2fb9e9208483e8402a8820b744241875ae97c8b2cc0fdcff1b307ac9e7a64 tasks/data_persona_aligned_multi_turn_50_0024/verify_workplace.py
789
  272a826b3fd8d66d30d336080fc16fcc48895fb8bb7eb45efa013a5e893063e7 tasks/data_persona_aligned_multi_turn_50_0024.yaml
790
  2d4b64c0f1268d3990b9bd5b89b7109881b13535cd1446567f697640a908a06a tasks/data_persona_aligned_multi_turn_50_0025/_env_builder_impl.py
791
  362812e444f590b5082c96d2d2549d64a1d1bf614fa3f3b24bb91525199c3062 tasks/data_persona_aligned_multi_turn_50_0025/env_builder.py
792
+ 698aafaaf414fefad8c36c9f5aa3934a553ce14f290e90a9d11701a4aca468ca tasks/data_persona_aligned_multi_turn_50_0025/verify_workplace.py
793
  b1f40c5349625101cfde2b597db7d8cc34c5e42c6c3caabc902115dbd19ed6e0 tasks/data_persona_aligned_multi_turn_50_0025.yaml
794
  6c463393ff0f0e82ee82f363cf49baf979dc1fa9033b4947396b222619c347c1 tasks/data_persona_aligned_multi_turn_50_0026/_env_builder_impl.py
795
  6d252c324058ded95411c3b8413678834a2342ecde91ec1b34bd8a44fe3918a8 tasks/data_persona_aligned_multi_turn_50_0026/env_builder.py
796
+ 14f2153f4eed27915b8f321b23489cd06cf08124578f5a102839516a98413dd3 tasks/data_persona_aligned_multi_turn_50_0026/verify_workplace.py
797
  b442abf8cb9ee1102004f826e795e06bad43f06c5d7bde8027e7e9b1d78d90ff tasks/data_persona_aligned_multi_turn_50_0026.yaml
798
  c0006cbabe3fdb665f6ec3447b20731e04b54f661d374135ec1a49ac4422f76f tasks/data_persona_aligned_multi_turn_50_0027/_env_builder_impl.py
799
  d9c5cfa216c64aab5a0aaa0c743209ecbaaee7b21c208b7be1bfa5187d5c596c tasks/data_persona_aligned_multi_turn_50_0027/env_builder.py
800
+ 4307b8a4a09c7a871846cfd9d008d674bae6f3ef37642abee700aa693ec14729 tasks/data_persona_aligned_multi_turn_50_0027/verify_workplace.py
801
  44311badecf8a14f9922783009223547d2906fe3c36a785555037480e4013d00 tasks/data_persona_aligned_multi_turn_50_0027.yaml
802
  060264e3bf2aff0a3968301b7a58b5d22ac0ba2de099fd51c7bd6899399712da tasks/data_persona_aligned_multi_turn_50_0028/_env_builder_impl.py
803
  1309ac0d0d1c41bf3c49be4fbebf6498c370653415c3c04582a746e70cc95aa6 tasks/data_persona_aligned_multi_turn_50_0028/env_builder.py
804
+ 293676cf396174c4a9512b7b2e0202920c5c665b1aead37aeec54ec033b935ff tasks/data_persona_aligned_multi_turn_50_0028/verify_workplace.py
805
  a3b8b7dbb01682d4e43b260fcf37e0e7667556fa29bba9e07d8544cef22800da tasks/data_persona_aligned_multi_turn_50_0028.yaml
806
  195613c7a4246e3cb5b333388a1cc4676097b9eb5ab5bdcba971acca63b3a7c0 tasks/data_persona_aligned_multi_turn_50_0029/_env_builder_impl.py
807
  97fc23c9048c1c2831a9101ed991365886af0dea36ba89bf44b69df8c9b0a413 tasks/data_persona_aligned_multi_turn_50_0029/env_builder.py
808
+ e45da5984624e2c82b438ad37f459c5edc078d7d99d91b99c434881cc2b78dc4 tasks/data_persona_aligned_multi_turn_50_0029/verify_workplace.py
809
  9218446a456d501644cd5caa96b0c7ef5b80c84baea1b12cf14f78c9bed7b5f0 tasks/data_persona_aligned_multi_turn_50_0029.yaml
810
  24c92dac93cd92431a861729d9f458e17ccc7df27598a58ba06b5add1446e4e5 tasks/data_persona_aligned_multi_turn_50_0030/_env_builder_impl.py
811
  712a694746bc6018a9c34b9c9c44ed9b47016d2a021faa98f1dc0e88ee723cfc tasks/data_persona_aligned_multi_turn_50_0030/env_builder.py
812
+ bdd74206422946c28e04a806a942e4d4091e3f861353ea0febcb44e7b1ef6f53 tasks/data_persona_aligned_multi_turn_50_0030/verify_workplace.py
813
  982f6d3ee34b2b08c5fed5b60bb4efa07276e7c672ccec891066334330e9936b tasks/data_persona_aligned_multi_turn_50_0030.yaml
814
  18e2d97bb019c59ab18a075164511385e25b922d32de046ff147ba58c4cd42b6 tasks/data_persona_aligned_multi_turn_50_0031/_env_builder_impl.py
815
  580eb15a196e6e37cc128e088dfe4ca6c1f17a6f52e42c5b60ce3a583d3f7566 tasks/data_persona_aligned_multi_turn_50_0031/env_builder.py
816
+ 18954b04bf6793561b1e01923a6164c4ec03de214d209ec6efd7b844e03b8471 tasks/data_persona_aligned_multi_turn_50_0031/verify_workplace.py
817
  00acac9c0ec49309b8c53bb84f43bacb1457a35c26da794dc6d1c7e1d8b3a8b9 tasks/data_persona_aligned_multi_turn_50_0031.yaml
818
  4f7e0c6b7d0085f1e41d1b0d8d734a8bb047fd9a27013a1c98ce559c73cfaabc tasks/data_persona_aligned_multi_turn_50_0032/_env_builder_impl.py
819
  38bd56ae9faf3307e0c61e4ce5369024f9ad7988c5008c2c23817e582bba4cb2 tasks/data_persona_aligned_multi_turn_50_0032/env_builder.py
820
+ c25c969fd2bbaebbccdc31e9e3da0164517857a959d643f82c9c92f132bd9abf tasks/data_persona_aligned_multi_turn_50_0032/verify_workplace.py
821
  fab672100494037ff01f33a25ece72b14d6123c7bfd3541ea7d5b85cd72d6b01 tasks/data_persona_aligned_multi_turn_50_0032.yaml
822
  f14ffd8e29bc23f7005434dd6afb2837f8a92826f9a22982cabb536489196cac tasks/data_persona_aligned_multi_turn_50_0033/_env_builder_impl.py
823
  8ac1faf9e5f109e1e6cd9ca568a343d5f4d29b865787f99ad5071eaeabce1826 tasks/data_persona_aligned_multi_turn_50_0033/env_builder.py
824
+ e64b7d4e6112dcc470a52db8107f183cbdd4a0b29727fa45d97ac3bd6a934d05 tasks/data_persona_aligned_multi_turn_50_0033/verify_workplace.py
825
  12711b324c0e31dced39daf7ac9addddaa7bc8a78032195b383f5477d320dfc4 tasks/data_persona_aligned_multi_turn_50_0033.yaml
826
  bbbe53f28b848f7781ad8954802120729539fb83886aaaf460e234d2fbb5a7ac tasks/data_persona_aligned_multi_turn_50_0034/_env_builder_impl.py
827
  0066b33da310f06a5a083507444277ba79af6664ecbe0f75d4d53a8e5412fc15 tasks/data_persona_aligned_multi_turn_50_0034/env_builder.py
828
+ 1edbff02132fdabbe44d16559c7e1e806f27fa23aa0d2c183a8d9a6ccbbe4766 tasks/data_persona_aligned_multi_turn_50_0034/verify_workplace.py
829
  52d961bd0000bd9cc7bcbeb74e5d2c99dbe321cb0973f66553ef8f7e17386b8c tasks/data_persona_aligned_multi_turn_50_0034.yaml
830
  5ae81d9ae4a23019537057c9a04c05d30daef5fa555afeeb690620cb2e956b0a tasks/data_persona_aligned_multi_turn_50_0035/_env_builder_impl.py
831
  83b535bccea97a26cf2fbf92a5611ed09d653371a76f32dff39ca89deaff03d7 tasks/data_persona_aligned_multi_turn_50_0035/env_builder.py
832
+ 0102a5c7a2980a067a1783bc92eeb04b7905cd808c54242f1f6f34ef6e794fe8 tasks/data_persona_aligned_multi_turn_50_0035/verify_workplace.py
833
  5e5801d61ce1eb2ccbeb4a045e0535298ebd7ded6bd82c334326ce9a57d8739a tasks/data_persona_aligned_multi_turn_50_0035.yaml
834
  c8edcd299904fb731fa74f8edf5ab980cf20e9ab6af8b33de4651c06c8afdfb2 tasks/data_persona_aligned_multi_turn_50_0036/_env_builder_impl.py
835
  0841055ba97d87d67fdcb56825f3b9b69a720687ae728d9fc73c70d428adc32e tasks/data_persona_aligned_multi_turn_50_0036/env_builder.py
836
+ 251e6c445d0f8a0a76e6ef70a822a480d136accd626b0d77d7032aa980bf0137 tasks/data_persona_aligned_multi_turn_50_0036/verify_workplace.py
837
  2ab96542a793e709c6a314e11edc7fe6118d4669ccfd69771b5fb37d194c828f tasks/data_persona_aligned_multi_turn_50_0036.yaml
838
  8e9ae3552c7b3cdf4fd1b743169708de0015351ecad1ceeba33dc220d82f5833 tasks/data_persona_aligned_multi_turn_50_0037/_env_builder_impl.py
839
  74c8ecd7d7eb79ec1816e362a855fc03c630d9394e80d0a522f850165052b500 tasks/data_persona_aligned_multi_turn_50_0037/env_builder.py
840
+ 5aa3196edbbf702485b25283bf0b40098f08056c7ad5605f0b8b4f963ee7774c tasks/data_persona_aligned_multi_turn_50_0037/verify_workplace.py
841
  871eb76552d080398c78a7cec0e4b052d1346a812aa617496e9d28f323771f92 tasks/data_persona_aligned_multi_turn_50_0037.yaml
842
  f42b5dc110a6237bf594811fba9bbd1e11c0140b9bf7123d1bc51e980b7ed502 tasks/data_persona_aligned_multi_turn_50_0038/_env_builder_impl.py
843
  871c46ddcd3c0cc59161027154f797fce686b2f5a5c18439f2459b97d1f2353d tasks/data_persona_aligned_multi_turn_50_0038/env_builder.py
844
+ ee4e2fdae62fd821c325681b2038c419efe12e6d9d6e156e223486988f77fc2f tasks/data_persona_aligned_multi_turn_50_0038/verify_workplace.py
845
  7d959df4fe308e26b221121fd0515965c21baf952e6188329bcd7f154f713b33 tasks/data_persona_aligned_multi_turn_50_0038.yaml
846
  a358a8ccc640abf6f5ef8c16e4c92cefd4a4b1d60f893d4a8a8693dc21c43596 tasks/data_persona_aligned_multi_turn_50_0039/_env_builder_impl.py
847
  f03b13c0760e04f97ec5afe38099a885868cfd26d8dd4d689e797086abf6eff1 tasks/data_persona_aligned_multi_turn_50_0039/env_builder.py
848
+ a968399da311e449cbeffbcb90248dbcd01df73b1a26e770f15a973b6ca9de6f tasks/data_persona_aligned_multi_turn_50_0039/verify_workplace.py
849
  5fd9a8dfa951f82891bf336194bad07cd698ac839613be821416214abdc45b35 tasks/data_persona_aligned_multi_turn_50_0039.yaml
850
  f6702cc7ca4c9a4c7766b3c19a648b0cdfb21684e0cc874e3eeba7d1b814b5cf tasks/data_persona_aligned_multi_turn_50_0040/_env_builder_impl.py
851
  40a157919ea9f49fc56c73354ec53675312fdd40a221d32cfaafeee5a4542a6d tasks/data_persona_aligned_multi_turn_50_0040/env_builder.py
852
+ ea3536a43b6a41f6592794560b3d9bb7d9d5cf15f249b07035c15233b961789b tasks/data_persona_aligned_multi_turn_50_0040/verify_workplace.py
853
  88fa960bc70839336fc4f4ef6dc4be347c91806648d02987c9dc3c6960a7d62c tasks/data_persona_aligned_multi_turn_50_0040.yaml
854
  226db282102ffde0d3285e4f5dba5f12582b6cb40e7575f71fdd80e55ac0d8fb tasks/data_persona_aligned_multi_turn_50_0041/_env_builder_impl.py
855
  bd0adce0018d723a97b587c08191e4561c98807881b8de5533df0c7ac62c3689 tasks/data_persona_aligned_multi_turn_50_0041/env_builder.py
856
+ a8d441a5369180690f015643115ed5ff7709e4ce18e014a888920bf286bba536 tasks/data_persona_aligned_multi_turn_50_0041/verify_workplace.py
857
  b7830cda94a157e8ee2152545ac021b7b030b91883ff6e67ee8c32fdd1b07ece tasks/data_persona_aligned_multi_turn_50_0041.yaml
858
  b826b95b287d0c441fc2f8943f25e783203fd14d0457e93684599c507956b801 tasks/data_persona_aligned_multi_turn_50_0042/_env_builder_impl.py
859
  80754de9f54c53cee72e4121334de887ee92b79d7a967a3b86cf985e33b2f6f0 tasks/data_persona_aligned_multi_turn_50_0042/env_builder.py
860
+ 1e0db16d30289caed2708d66c50f9e462b32d281c0b96398485c8df43ceea100 tasks/data_persona_aligned_multi_turn_50_0042/verify_workplace.py
861
  e97fe1782faaacc41fe8709b59c71fd3e61ca30098c21ea7bef75d5319433c23 tasks/data_persona_aligned_multi_turn_50_0042.yaml
862
  bc94a1a9d8edce537f51e3dbcdb56c5cd5c44d61effd937eb4e9b142f26e539d tasks/data_persona_aligned_multi_turn_50_0043/_env_builder_impl.py
863
  4e8977b9be57fc42568674d2664e5ca27569bb49b6ea075a7cf024ea88de70dc tasks/data_persona_aligned_multi_turn_50_0043/env_builder.py
864
+ a9e569b02b2a308e4a234815ddc7c2f898ecef4e7ef0678b71bb069031a1555b tasks/data_persona_aligned_multi_turn_50_0043/verify_workplace.py
865
  cfde3128cf17843380a7d244ba6dd2d3113594a8c35bd2a07b73716504a1e237 tasks/data_persona_aligned_multi_turn_50_0043.yaml
866
  c7846fb438cb46bafbd8b73427e89cf9ec898258802996231e06a034e80cbc62 tasks/data_persona_aligned_multi_turn_50_0044/_env_builder_impl.py
867
  fc215986f1c8e4d29d5474315f49b76e227d5aaf095bc23304bd0345574cf07e tasks/data_persona_aligned_multi_turn_50_0044/env_builder.py
868
+ 575fda1920ff7108634bce3574b52cfc907b78fda7e9a5f99279817a553d8d84 tasks/data_persona_aligned_multi_turn_50_0044/verify_workplace.py
869
  48b45df848822f760f65c743e77f425971c4af190631be7e9a218c5cc83f8f3e tasks/data_persona_aligned_multi_turn_50_0044.yaml
870
  3a4e143a8ad078eb251ca69de34ba587c270b7b6651060823cc34afe9d264e4c tasks/data_persona_aligned_multi_turn_50_0045/_env_builder_impl.py
871
  0d51165aca6030bd0d8360e05660e6a8cd2b8351cadd0a89054551b9ea0244bc tasks/data_persona_aligned_multi_turn_50_0045/env_builder.py
872
+ ddaa328e7c20f38f8e92fd7cc06de86e0c3f4b2c88c3efb21f9789109f25744a tasks/data_persona_aligned_multi_turn_50_0045/verify_workplace.py
873
  9d6cee36d66dc5afb78db5980b1cbcec9cbbb21cb1b6597764bf7d308acd8616 tasks/data_persona_aligned_multi_turn_50_0045.yaml
874
  ff492138561fbc06ac80201ee47e004adca3cfc4187d35dabdad7a75889a0ec7 tasks/data_persona_aligned_multi_turn_50_0046/_env_builder_impl.py
875
  1c3ba59afd76b331fe62b35cb6d4a0efdd04e69e4910ebd3cf0067d4fe765b75 tasks/data_persona_aligned_multi_turn_50_0046/env_builder.py
876
+ 8f8aaac293f11031cd6288fc3661d388a3f09fd24e8d9f75ea5b67c65b9b1739 tasks/data_persona_aligned_multi_turn_50_0046/verify_workplace.py
877
  dc011dd019bc9673cb5916799b91bd9c9145571f009fcf34657b2517fda83a09 tasks/data_persona_aligned_multi_turn_50_0046.yaml
878
  5b5ed722e3df8926facbc4f6620b397d8e7e3fb3a605e46877111831485d0868 tasks/data_persona_aligned_multi_turn_50_0047/_env_builder_impl.py
879
  32f976d9480f4c2775ecdcfb3b875985fbc65f3ca49337bfd43f2c4696f8258b tasks/data_persona_aligned_multi_turn_50_0047/env_builder.py
880
+ db9042d57fd54fc326acb68e1d377bc0bb723652b144ab77577a8d50a0be4c96 tasks/data_persona_aligned_multi_turn_50_0047/verify_workplace.py
881
  c3f617ca33e27dcca0fe41dcd601b8b1e5d3c11325de1ca8f31a969307d8bffd tasks/data_persona_aligned_multi_turn_50_0047.yaml
882
  09f26a5bf64800014c2f0244624998f97df1e39dbbd4c85555f6b6c8b7abda8c tasks/data_persona_aligned_multi_turn_50_0048/_env_builder_impl.py
883
  01d5378944c09344f9c3627ecfb90a633b5cdb76ff858d6470ac2c4e51414375 tasks/data_persona_aligned_multi_turn_50_0048/env_builder.py
884
+ b07026fef46469553001ff1f45c7c3f88211f8167d1cb96892f1934f2b2bdafc tasks/data_persona_aligned_multi_turn_50_0048/verify_workplace.py
885
  d58c989e4fe9248e102558deb63fc15c0084fcbdd0ee4b0b01d77a0bcff78562 tasks/data_persona_aligned_multi_turn_50_0048.yaml
886
  21c8013efb0e96630bc2c90040832c130e4208462d8c0ba80119af6afacffd4a tasks/data_persona_aligned_multi_turn_50_0049/_env_builder_impl.py
887
  4fac6acb9722be1a134dd388f6a895bfb17e52244ee15d6282fa71f8253dbe6e tasks/data_persona_aligned_multi_turn_50_0049/env_builder.py
888
+ 28f9f8767014e00d60d71c8fa433db7b26edd5fb73c3edee2387c1d4d0ad1315 tasks/data_persona_aligned_multi_turn_50_0049/verify_workplace.py
889
  644ddefeb0fbee316e2f60b216062ef146f90b1a67b138051e30b005561bbced tasks/data_persona_aligned_multi_turn_50_0049.yaml
890
  b191844585401540c79229e85dcbcb61d3229d50c24f77cf22620e30c3b83ee1 tasks/data_persona_aligned_multi_turn_50_0050/_env_builder_impl.py
891
  71d53a790ea3c2a3cd3f2168b934cc8fa3d25acc49dc3831c2ab0a9da1663b1f tasks/data_persona_aligned_multi_turn_50_0050/env_builder.py
892
+ a8937dc97d67b5784ba57a28caea1e7a100a24576abe044cde74f8c50df9292f tasks/data_persona_aligned_multi_turn_50_0050/verify_workplace.py
893
  d4b7b8462bf456d77f9914c931edb42a1f03ca9de288a3caf7a72c2414cfcc2a tasks/data_persona_aligned_multi_turn_50_0050.yaml
894
  9ba5001c18264e7492816c699544e57c474ac01319b5199ec419f613dd5d236a tasks/data_persona_aligned_skills_50_0001/_env_builder_impl.py
895
  430f0da57468b56d4d7cd5d665c3c4233d8fd9c16ad71c99f2d44a88fc2d5b59 tasks/data_persona_aligned_skills_50_0001/env_builder.py
896
+ aa66d949e6b0ece5dccdc6da6f90345a114893b7feed3ef0397289addac34531 tasks/data_persona_aligned_skills_50_0001/verify_workplace.py
897
  4af3e9203475eeea4f938b3afc4b9e720008b850b88561dcb8fee7d3bab40bfd tasks/data_persona_aligned_skills_50_0001.yaml
898
  fc24bbeda0fdfbc36b8f1af148470a2d5aa53bd49cbcde7bbc69fd3d2e662351 tasks/data_persona_aligned_skills_50_0002/_env_builder_impl.py
899
  75a883f3da5b76f3f3dc087ed2af67352131c55627aa8f1bb4bc0bcd4ecedd71 tasks/data_persona_aligned_skills_50_0002/env_builder.py
900
+ da05c135493a8a7a13625ff52dd3f4a80637ed85d4dcbdafe251a1bc3be7b5af tasks/data_persona_aligned_skills_50_0002/verify_workplace.py
901
  1b0ec155c383fb6e1132bf92c5abaf53349ca5119e232db1a0abc18400936cd0 tasks/data_persona_aligned_skills_50_0002.yaml
902
  bfa32aa5d8614b836be8c879484034a8d62dad89e85578450f5adfb2a9ab88f1 tasks/data_persona_aligned_skills_50_0003/_env_builder_impl.py
903
  53d248a85bb661a1cc67b1f96fbf9ea638ab2b6556d337f23e5a432f794b62e6 tasks/data_persona_aligned_skills_50_0003/env_builder.py
904
+ a2e8e040770cc6589f0defebba963785470c10ca1bdea119303f152ec9258d40 tasks/data_persona_aligned_skills_50_0003/verify_workplace.py
905
  885f054d78faf097189ee04b97fe326a7944218babce1175ec330352de7dde24 tasks/data_persona_aligned_skills_50_0003.yaml
906
  d104f49cecdae66d6a664bc99627dda16f416a6864b398dc224d76ccffde78ea tasks/data_persona_aligned_skills_50_0004/_env_builder_impl.py
907
  0d3147dfb30a7098c44f12446ea2fa2da39af6b68c9d00775b28cb2d9217cd49 tasks/data_persona_aligned_skills_50_0004/env_builder.py
 
909
  2e41276eb8dec0024c82210103919dfc64880d3af4af90a10172a2aeede57072 tasks/data_persona_aligned_skills_50_0004.yaml
910
  f0265e16e5276519ba1562ef3c356132b5cd8cbdae845e7c0423c8fdcb80e7b7 tasks/data_persona_aligned_skills_50_0005/_env_builder_impl.py
911
  5ea44bc46399d477cc8ca7bc611eed0992cb0ff05fe4cc40e175362389ae41db tasks/data_persona_aligned_skills_50_0005/env_builder.py
912
+ 2a9b778cecd08e6132f152250d1f1aba7d57bf5571655d675ff6fd1f96583810 tasks/data_persona_aligned_skills_50_0005/verify_workplace.py
913
  b63105179c6ed7ee525a902045ac208c3f9a51df503ee9eaf708429bfb4c742b tasks/data_persona_aligned_skills_50_0005.yaml
914
  6c24e33be7b534bc2f825b514b962335d5dee4ca6b8329af9da055993bf92072 tasks/data_persona_aligned_skills_50_0006/_env_builder_impl.py
915
  344f2bc4bfcc1b8231fe15c7ff2b5a30f343ad02527c6c90f02904f8fd027dab tasks/data_persona_aligned_skills_50_0006/env_builder.py
 
933
  ce5af4d9f8f8cce1cffac88c30be0075ced9d39e3288f35572041a1172a12588 tasks/data_persona_aligned_skills_50_0010.yaml
934
  7c825463e343c1f07c463fb265309a6ee73fa3c2868230a41072e806a0d386c5 tasks/data_persona_aligned_skills_50_0011/_env_builder_impl.py
935
  737637e643d3c9e11ce2de1b00d0d643947020efdc8f217281f67acef6ddfebd tasks/data_persona_aligned_skills_50_0011/env_builder.py
936
+ 1f2764916e1a21112902de6cdccd5630f3d580d440d7ff6444717a7262303060 tasks/data_persona_aligned_skills_50_0011/verify_workplace.py
937
  fe955467d70c0c91443b7cbbf1af4652139c58edeee259ae5a0d1a57ff1b1d34 tasks/data_persona_aligned_skills_50_0011.yaml
938
  2e4c21330127825cd38bb129db319ad3da984675957158ea76755166b6d69dc6 tasks/data_persona_aligned_skills_50_0012/_env_builder_impl.py
939
  e179c7026efddd0374bc6f841d4ba8bf44e809af7feab3e5f8a52db7f64676d1 tasks/data_persona_aligned_skills_50_0012/env_builder.py
 
945
  08b46e279e9768961e7099589b3fabfcd49f9f68f764234ba2426bb4a8c75944 tasks/data_persona_aligned_skills_50_0013.yaml
946
  8c94e13b8c2e270164732c96ed73b108dc961da4c6d2c7e02c7b2807935960f9 tasks/data_persona_aligned_skills_50_0014/_env_builder_impl.py
947
  c736803229566964a42a2eb17ec13d79bd106447f1a31890cbe22a761df25e65 tasks/data_persona_aligned_skills_50_0014/env_builder.py
948
+ 988369039f8360f3093ef18a7f931746c402dd40fe91b4d81743f9553bb4f6b8 tasks/data_persona_aligned_skills_50_0014/verify_workplace.py
949
  4f04eb44f689489ea3a7733917c527d679adad8e11dcab3e1449c016b764b17c tasks/data_persona_aligned_skills_50_0014.yaml
950
  354f87b22d9189ad51894259d7ff75bb026d010a72e660584bb326f004082ad8 tasks/data_persona_aligned_skills_50_0015/_env_builder_impl.py
951
  aebcca1c91e3c94958570841f26677ce36520b1fe4245cb02bd50f46576a81c2 tasks/data_persona_aligned_skills_50_0015/env_builder.py
952
+ 1ebd329fc3dffed8887b8502ec72aa87cb3ba145ee65138fe69c950fdeccfa4c tasks/data_persona_aligned_skills_50_0015/verify_workplace.py
953
  7a27461aa88ba0eee05a784d2a47012e839b3e5ec9c5e4f111aaad6ef7b20123 tasks/data_persona_aligned_skills_50_0015.yaml
954
  cc779485fe3eb4f49c1e7f813deb9296b8e9d859ae5385dc74dc713b637a458a tasks/data_persona_aligned_skills_50_0016/_env_builder_impl.py
955
  c333538022d2e41613ae1bf79dae2917581c5d09a19f38fcb2f20781844b6281 tasks/data_persona_aligned_skills_50_0016/env_builder.py
956
+ b5ade2c5f8ca5e984edfcd762f84a499dc2650946c225cca7c2461aaaf7d7a8a tasks/data_persona_aligned_skills_50_0016/verify_workplace.py
957
  fe8153cfb7818ca1a90518bbf1d7e87f615e3b7fb57b4c0d1a188c94fb0b800a tasks/data_persona_aligned_skills_50_0016.yaml
958
  b9911a24f4c2daacc3607ed812e58abc927d8b3c1b0b32ad82294352141e947e tasks/data_persona_aligned_skills_50_0017/_env_builder_impl.py
959
  60f0d4e7033f8216a46abf7cd841b23b2222528c40857630c868a17f58df5fcd tasks/data_persona_aligned_skills_50_0017/env_builder.py
 
961
  8326b7b45697e7c849bcf5b52a1aef8efe8c82b7d8be56151b08d4d002a67f31 tasks/data_persona_aligned_skills_50_0017.yaml
962
  521c60bbf958efb0e7ea8b38c32d2f617ecad0f09b460cc8313dec307499fb3e tasks/data_persona_aligned_skills_50_0018/_env_builder_impl.py
963
  d731a7ec2ab7ccf43db400b90e85646c3342a2a614adcdc225fe8c5cd7bcc50a tasks/data_persona_aligned_skills_50_0018/env_builder.py
964
+ 09ca555307c02ba65db8c2aa2ceb9c60fc55b58a78d9cf7a4aff61c7c5d77213 tasks/data_persona_aligned_skills_50_0018/verify_workplace.py
965
  e873b6c140b7b534715ca2ceac136fe62591be2cc40503e139d504b9e82b02b0 tasks/data_persona_aligned_skills_50_0018.yaml
966
  d9805c03d587b088b94ccccae6560b01fa66433ca2e441276e091887340f4015 tasks/data_persona_aligned_skills_50_0019/_env_builder_impl.py
967
  fa6755f08d5e6f0d9d4f35271d7fe03042787656ac4bbbfe3444c505905473f2 tasks/data_persona_aligned_skills_50_0019/env_builder.py
 
973
  fabd3cf3944965bb4f516774215504241e9e3db295e73a5ceb4586668d595bba tasks/data_persona_aligned_skills_50_0020.yaml
974
  ed9434d766f1a2e81a0daae2cfdb59820535b44e852ef1dec0d051869ae4cf59 tasks/data_persona_aligned_skills_50_0021/_env_builder_impl.py
975
  1521cab300a19371492873f2bb15ed83a2782e98109cb5e89d04043820b50694 tasks/data_persona_aligned_skills_50_0021/env_builder.py
976
+ b3744006d9b04e9015f77b6190ba878f6139cbabe1ddbd15056fb0fe0ab9da6e tasks/data_persona_aligned_skills_50_0021/verify_workplace.py
977
  87910d80f20fd04a2ef55901b74a011838ecaa511b3bc9366d9e5d1638c09df7 tasks/data_persona_aligned_skills_50_0021.yaml
978
  06bd01da0800305da5427aa4d1c130ec5b0760ca8ff82f62ede66efa45069b1f tasks/data_persona_aligned_skills_50_0022/_env_builder_impl.py
979
  0e56b3c683effb4992409bc0056e7936c97ec27d7719d414b08175883ab028dd tasks/data_persona_aligned_skills_50_0022/env_builder.py
980
+ 83311ef366ece854ccc7e16a3d75367b48472a8bb0f3a370137518d12c94e493 tasks/data_persona_aligned_skills_50_0022/verify_workplace.py
981
  520535d3f238adc3005c90702a47300ad23b24fb84a2e411ccd2d6340115c6e1 tasks/data_persona_aligned_skills_50_0022.yaml
982
  e31d518422ef198098f962ba3ce48bf8249f418a3a32075093fa5b2d75803bed tasks/data_persona_aligned_skills_50_0023/_env_builder_impl.py
983
  563dcc2509374b1a44788adfd8d17be421c2a8b77e5b2d6a2a82d4cbf2e29ad2 tasks/data_persona_aligned_skills_50_0023/env_builder.py
 
985
  4bad7c53510de37ff6813ce69fbc12d290c1ad6a3cd4cbfd1d6de0fe3244c626 tasks/data_persona_aligned_skills_50_0023.yaml
986
  d35963fd45280725c1571f25d1d08fc82cfe2c7078b7172f235f73afd31d8d02 tasks/data_persona_aligned_skills_50_0024/_env_builder_impl.py
987
  8799c3bdfa812fd2986a50ebda72389e9a5aeca523e9f946cc0557afa3d0b36c tasks/data_persona_aligned_skills_50_0024/env_builder.py
988
+ 9d2e9ebf5b3d50dffcf4be0597517b81bb81a23e79302fa889e3c657f737f2c2 tasks/data_persona_aligned_skills_50_0024/verify_workplace.py
989
  052e79a68e27e08178852a47d678a4da24baa9bd3434c20c5247c282bbdd9f1a tasks/data_persona_aligned_skills_50_0024.yaml
990
  1efafc6fb51758486d20100fd773c07a991507c921dbb94b0e452be8119faf60 tasks/data_persona_aligned_skills_50_0025/_env_builder_impl.py
991
  229064e628c13e26467710700e915658c662c7b346553272384169f31ce90f7e tasks/data_persona_aligned_skills_50_0025/env_builder.py
 
993
  7737c2b596b3d610b399a9f33a32e46fc4fecf89204ae3e2a64545a764f5afaf tasks/data_persona_aligned_skills_50_0025.yaml
994
  3c1bec3f4470b12e5d77dbd90a24ab2a76154e8f9a7830774549252db0ac4cd3 tasks/data_persona_aligned_skills_50_0026/_env_builder_impl.py
995
  933416369c3bb58ba3314503c887c6c8f577e6b737f283a4f36ac119913911cb tasks/data_persona_aligned_skills_50_0026/env_builder.py
996
+ e94d7d5e5832a5c0a6fbd8308faeb4162b0d8ee230d7f06320c5b425cab16284 tasks/data_persona_aligned_skills_50_0026/verify_workplace.py
997
  9c8cdb25f86d02bc56ce53359ecad12fb088425ae74ae4f0cdc4a728d0342457 tasks/data_persona_aligned_skills_50_0026.yaml
998
  53176ff4bb81e872166aa77535ea05416588fc365381b1b6fe18921468227d13 tasks/data_persona_aligned_skills_50_0027/_env_builder_impl.py
999
  ffa0c07465ab60eadad5fe2bae3013d5df6ea4e0291e72df7ebcef4ce378a0ea tasks/data_persona_aligned_skills_50_0027/env_builder.py
 
1005
  b512a128cdce54330a0f593481fa9c49aeee361ad5fb2c752ff030f187fe3543 tasks/data_persona_aligned_skills_50_0028.yaml
1006
  ef11a48e0bff07f41db212d1e056641e6fb84bf7092d4de9f1cb12c89aac963e tasks/data_persona_aligned_skills_50_0029/_env_builder_impl.py
1007
  9147038f21e5c68c2f58069a4a80d1f69400ebe6bc7b5c745e122d32461073aa tasks/data_persona_aligned_skills_50_0029/env_builder.py
1008
+ 154a3e2a7e98c4f6859a116177177def9759a4798c27e324c47c457b554f8022 tasks/data_persona_aligned_skills_50_0029/verify_workplace.py
1009
  ac86eb2860d2e0d9748d160e6b65a6b63c9df1013b9b983d54f81fdfe859599d tasks/data_persona_aligned_skills_50_0029.yaml
1010
  01b4a62c7aa5471bd101e9cff336caab0e205526f20e85068e33281b15caa14f tasks/data_persona_aligned_skills_50_0030/_env_builder_impl.py
1011
  ac362980bbf9cfb0af61bdb5ee9e393efd270650ff9c3754bcb1e5c16390b00e tasks/data_persona_aligned_skills_50_0030/env_builder.py
 
1021
  c335a6527dd41dc3305842d835c57e1adae5adb7b27c99ab6ca82927d458aad4 tasks/data_persona_aligned_skills_50_0032.yaml
1022
  ae6beef9ce164ded168785e5673e9d5ef52d2a58873256107e721f0df07d2d05 tasks/data_persona_aligned_skills_50_0033/_env_builder_impl.py
1023
  cf492a14d6ba8cb1389d6553ee3922d1d43d8e632a80f5bba1c49c8cd8841385 tasks/data_persona_aligned_skills_50_0033/env_builder.py
1024
+ 67ca834ab57dbf2dff35d911c2c6f4b392fa2730c7dc8706a06a9c2b86cd24cd tasks/data_persona_aligned_skills_50_0033/verify_workplace.py
1025
  feabc1f80335bc3347ec9c98941e726c6d0d9c80c1687471cd078c39679a2b7c tasks/data_persona_aligned_skills_50_0033.yaml
1026
  c9894b735720a06bb0437dd724047e679ab9713cf7789cdbdfea17beb19a30b7 tasks/data_persona_aligned_skills_50_0034/_env_builder_impl.py
1027
  ee0aa6e61f651e79bb533bc0cd32a830f01efd9c5aada1cbd35ccdc89f74d404 tasks/data_persona_aligned_skills_50_0034/env_builder.py
1028
+ aaf4b59cd0fa66f9c35577561786b3f1780bba6440c6071f2521d1721e33b446 tasks/data_persona_aligned_skills_50_0034/verify_workplace.py
1029
  0427163956b3de9af9a5424289a4cce349e5538996f1f01b02bb1679bec12860 tasks/data_persona_aligned_skills_50_0034.yaml
1030
  9bb3d0e5a3664de1334501d49ea7259f3cae0acf77bb3214f4d765c4023b96a3 tasks/data_persona_aligned_skills_50_0035/_env_builder_impl.py
1031
  85c680d11531ff47dab821ace681e14aaa7f07074c471116af8225a0064a6638 tasks/data_persona_aligned_skills_50_0035/env_builder.py
1032
+ f357da86f2272c55ed85aa970abfe976897ee2e03ec3094385d0c374aa8a45f6 tasks/data_persona_aligned_skills_50_0035/verify_workplace.py
1033
  19d53076922ece67067be7cb16488f31d02a81e3e50ba3fa981d18872ae92f71 tasks/data_persona_aligned_skills_50_0035.yaml
1034
  0bdad4710b0ac257843288c516f7403c65a06f28981d5faa9e0fda4248e0b014 tasks/data_persona_aligned_skills_50_0036/_env_builder_impl.py
1035
  5c00d5a4240ae379b108e153f893be0dc3bc4b07d020fa1f4bdd5d20aaede295 tasks/data_persona_aligned_skills_50_0036/env_builder.py
1036
+ fe1796d6d886fffd3406fad7b3de955942db39ae0362b627726264bad7d5d4a7 tasks/data_persona_aligned_skills_50_0036/verify_workplace.py
1037
  d225a42bfe56bf3c4a3d943ba4fb8e04aa3c2478adb18bb40fe52247a4b0c622 tasks/data_persona_aligned_skills_50_0036.yaml
1038
  22aad0a3550173e6745e1cd169614aeaa7e508ea2ceb7c69fc1f6ccd332844d3 tasks/data_persona_aligned_skills_50_0037/_env_builder_impl.py
1039
  feb57314ff7674cbb29a97127d8364fd72a2b9b1a6843f95e6365b9d8b532ecd tasks/data_persona_aligned_skills_50_0037/env_builder.py
1040
+ 397ff2ada1944fb3df335ee859ea7967ebb15732952617309660b164220790b0 tasks/data_persona_aligned_skills_50_0037/verify_workplace.py
1041
  7c9a74cdcc43bcb4d4f75e761a0dc3bb25604680f69533dd63bb4c00271665a2 tasks/data_persona_aligned_skills_50_0037.yaml
1042
  01001db221ce59adac2d4ba0c2b97efcd0b0c2488964009f833ae938936fc90f tasks/data_persona_aligned_skills_50_0038/_env_builder_impl.py
1043
  45d60213b88dd3693717dd38b83fb533ed6988bb7c24c14815cbfb9ee7543466 tasks/data_persona_aligned_skills_50_0038/env_builder.py
 
1057
  412d23bc38ad9f7aead3bb5f17d2c1637da3fa4193874c72afeb71c62350f67f tasks/data_persona_aligned_skills_50_0041.yaml
1058
  140c27a7bb83249c9e2298167294af52b8985628e87b31823c767d96f07c26ea tasks/data_persona_aligned_skills_50_0042/_env_builder_impl.py
1059
  ffbb8d53084d7fbc08206f9f75780c81246f5a4e014502fb4c067777068b1a8d tasks/data_persona_aligned_skills_50_0042/env_builder.py
1060
+ c3e65d1975a143372961741ee5f6e1da2de956883dca68f5623e5afb40a68a5d tasks/data_persona_aligned_skills_50_0042/verify_workplace.py
1061
  d230f00e94bc52eca65dd51512f96c1d6efeae851bb868c9c4d148edfc110190 tasks/data_persona_aligned_skills_50_0042.yaml
1062
  c075e9406486d573d4632833c728b4f60d218c14563c163faa6958c8095906a4 tasks/data_persona_aligned_skills_50_0043/_env_builder_impl.py
1063
  cbf0a4977e35a0e14fc9312d26992a6c2206ffed2ec7e4d7738b2495f53d0a6d tasks/data_persona_aligned_skills_50_0043/env_builder.py
 
1065
  773bbd1b442826108f6ba98877d759011a295d8353b0b84fb6897dbac748b6f3 tasks/data_persona_aligned_skills_50_0043.yaml
1066
  55b89991abffc6b0fd53b2660764cf9724aef3c90e75a00689fbe63c926acb81 tasks/data_persona_aligned_skills_50_0044/_env_builder_impl.py
1067
  6a01c6cfbf804d535758152d64ec149c68447a159a97c13ac4ac4ae0ef794e6f tasks/data_persona_aligned_skills_50_0044/env_builder.py
1068
+ ff7b2ea4d517c86797d32b5520591da0233606ef80dc413b4dbeed6d1efef0e1 tasks/data_persona_aligned_skills_50_0044/verify_workplace.py
1069
  72264c7d2c9337565af2dfdf6dd496c153e264715df0ac16724f0aef9b5fce37 tasks/data_persona_aligned_skills_50_0044.yaml
1070
  962cc304260ff512b2f2600fae9ca9e139fe6b738f78d5d10ac4faeb79e0d23e tasks/data_persona_aligned_skills_50_0045/_env_builder_impl.py
1071
  883b570204c8cf928af8390ec30822fda289c49b03eb0c9c38672bad4627ba6b tasks/data_persona_aligned_skills_50_0045/env_builder.py
1072
+ a04fc50f267c0616f29496ce04f10fb265427957de527f75e13c5afe126851ea tasks/data_persona_aligned_skills_50_0045/verify_workplace.py
1073
  87831e1000803c11f8aa843c009acb9111b4719565e07f191b35e3a868847dd3 tasks/data_persona_aligned_skills_50_0045.yaml
1074
  ca397f3f94e7132c5224262ee863cbe9f04dc5b3c82661e2603dd0c1f8058413 tasks/data_persona_aligned_skills_50_0046/_env_builder_impl.py
1075
  c6ccdc8c10e5aea49d7b5c2c2f924121eeb9f11a1953e9b4cf7bcd272128412b tasks/data_persona_aligned_skills_50_0046/env_builder.py
1076
+ cc779bf64d79c631ed267a05d6c98dd0576920bb752c9c30a0952ae400b9174d tasks/data_persona_aligned_skills_50_0046/verify_workplace.py
1077
  65e50bdafa9d8adf5ccd35aaf1570b7b21d3bc5fa81726c3e5090b349486e457 tasks/data_persona_aligned_skills_50_0046.yaml
1078
  d00ded6ebc6b3b6d62e6db06af2909d201e355b37cd6f9b730370a9d6cee2419 tasks/data_persona_aligned_skills_50_0047/_env_builder_impl.py
1079
  3ba4f6142e9e6719e30e69d4dd39b5335ef5d1a719af6bbf1bd989f5f7f0b7da tasks/data_persona_aligned_skills_50_0047/env_builder.py
1080
+ 0550ca6e85a9f657b236468ded733fc7ae0c599d3a76c389d0df95db8d0a4457 tasks/data_persona_aligned_skills_50_0047/verify_workplace.py
1081
  b22b0d5e24c65817e6c535d395cf20e9641b0b43f78e93a248f3a748cc0a40b8 tasks/data_persona_aligned_skills_50_0047.yaml
1082
  bd3211d73e553fb3a9d1c3ffad81baf1b715f6259a8e3d9bd742af362b763fa5 tasks/data_persona_aligned_skills_50_0048/_env_builder_impl.py
1083
  a0d75c1bec36c11ab36a64aabef358d67b5f5b37588e2761e330647ea54c1374 tasks/data_persona_aligned_skills_50_0048/env_builder.py
 
1089
  f078c6121ad45719dd02c2e90b1b7abb171665bf20dfac83337a4868743bf04a tasks/data_persona_aligned_skills_50_0049.yaml
1090
  1bafc5e8b098ca733508d5dd1a1986da3c60015aa57f2318a6b80a47028a4085 tasks/data_persona_aligned_skills_50_0050/_env_builder_impl.py
1091
  30158949469822f94fd21d511b6df006c3767f9130a0425a58750bc3bb856171 tasks/data_persona_aligned_skills_50_0050/env_builder.py
1092
+ eed5b9b52eb34c3cb84b197eb8a0b507bdae59dd6505a16196dfcda7ed2e8b8f tasks/data_persona_aligned_skills_50_0050/verify_workplace.py
1093
  bae66782f8c08b21ecba97700782f5ef09dc4b2c805d0705415ae6de41a7ca2d tasks/data_persona_aligned_skills_50_0050.yaml
1094
  9e65f31d924406cea0ae3ec57d5a174143672d11733f9480e0acf9d1ce1459fc tasks/prompts/data_persona_aligned_base_50_0001.md
1095
  0e92ca66c816d518f672fee4a64b357935c59b12f73c136a44a434d91331bae3 tasks/prompts/data_persona_aligned_base_50_0002.md
 
1391
  318e330f318d95cc36aab3076c495ddcab142476156d939c2038a4378a575aaa tasks/prompts/data_persona_aligned_skills_50_0048.md
1392
  93002945e85af62715331c5abbdd6509430ff1ae5ca8428f0ae1ede1f38624bd tasks/prompts/data_persona_aligned_skills_50_0049.md
1393
  539d7f3471c7a25cce4874b93d01f7a03cf162fbea98c16182fa8a838107f5f5 tasks/prompts/data_persona_aligned_skills_50_0050.md
1394
+ 8a376b90cd229312b8c9d5ba18400b378faff2ff21fd2c43626f8b212fe77954 verifiers/base.jsonl
1395
+ 2e7dc5bce993a11b664a367df0b791beb9cc3d95c5c9d1a4cd6bf0ff2621e2fe verifiers/hard.jsonl
1396
+ 70feceb0df9a8ba7a6f5a9e0d989d276adc10ee3a395a571353b0a327791589c verifiers/multi_turn.jsonl
1397
+ b3c78f66546199c4b2fdd3be2fd50b506fdcc8441c012d29936307a59ace19f5 verifiers/skills.jsonl
persona_aligned_mix_200/manifest.json CHANGED
@@ -39,8 +39,8 @@
39
  }
40
  },
41
  "files": {
42
- "count": 1397,
43
- "bytes": 6118979,
44
  "checksums": "checksums.sha256"
45
  },
46
  "skills": {
 
39
  }
40
  },
41
  "files": {
42
+ "count": 1398,
43
+ "bytes": 6266949,
44
  "checksums": "checksums.sha256"
45
  },
46
  "skills": {
persona_aligned_mix_200/provenance/verifier_materialization_manifest.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:216e7ba29a42c5bc74324bc1c0c536b5c2103a13f9eae7f05831598cd4e8257f
3
+ size 127213
persona_aligned_mix_200/provenance/verifier_repair_manifest.jsonl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:77c36f2054abdaf4ff577ee9dc1e04ab6352ce1761c1720aab69e95125b4a29d
3
- size 32165
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bb169e2015ddd2274fe010287da68bd2c3aeda72e4b1062adfa5445fc0b17260
3
+ size 33778
persona_aligned_mix_200/tasks/data_persona_aligned_base_50_0027/verify_workplace.py CHANGED
@@ -1,152 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import httpx
5
- from openai import OpenAI
6
-
7
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
8
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
9
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
10
 
11
- # 初始化客户端，必须关闭 SSL 验证
12
- http_client = httpx.Client(verify=False)
13
- client = OpenAI(
14
- api_key=MOCK_API_KEY,
15
- base_url=MOCK_API_BASE,
16
- http_client=http_client
17
- )
18
-
19
- def llm_judge_content(prompt_text, file_content):
20
- """用于检测非结构化文本的统一接口"""
21
- try:
22
- response = client.chat.completions.create(
23
- model=MOCK_MODEL_NAME,
24
- messages=[
25
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
26
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
27
- ],
28
- temperature=0
29
- )
30
- return "yes" in response.choices[0].message.content.strip().lower()
31
- except Exception as e:
32
- print(f"LLM API Error: {e}")
33
- return False
34
-
35
- def verify(workspace):
36
- details = []
37
- total_score = 0
38
-
39
- target_path = os.path.join(workspace, "recovery", "target.json")
40
-
41
- # 1. 验证结果文件是否存在 (10分)
42
- if os.path.isfile(target_path):
43
- details.append({"item": "检查目标文件 target.json 是否存在", "score": 10, "max_score": 10, "passed": True, "reason": "文件已存在于 recovery 目录中"})
44
- total_score += 10
45
- else:
46
- details.append({"item": "检查目标文件 target.json 是否存在", "score": 0, "max_score": 10, "passed": False, "reason": "未找到 target.json 文件"})
47
-
48
- data = None
49
- if os.path.isfile(target_path):
50
- try:
51
- with open(target_path, "r", encoding="utf-8") as f:
52
- data = json.load(f)
53
- except Exception as e:
54
- details.append({"item": "检查 JSON 格式合法性", "score": 0, "max_score": 15, "passed": False, "reason": f"文件不是合法的 JSON 格式，解析失败: {e}"})
55
-
56
- # 2. 验证 JSON 架构完整性及无多余捏造 (15分)
57
- if data is not None:
58
- if not isinstance(data, dict):
59
- details.append({"item": "检查 JSON 结构体", "score": 0, "max_score": 15, "passed": False, "reason": "JSON 的根节点必须是 Object/字典"})
60
- else:
61
- keys = set(data.keys())
62
- expected_keys = {"rank_id", "coordinates"}
63
- if keys == expected_keys:
64
- details.append({"item": "检查 JSON 字段完整且无多余", "score": 15, "max_score": 15, "passed": True, "reason": "包含且仅包含题目要求的 rank_id 和 coordinates"})
65
- total_score += 15
66
- elif expected_keys.issubset(keys):
67
- details.append({"item": "检查 JSON 字段完整且无多余", "score": 5, "max_score": 15, "passed": False, "reason": "包含所需字段，但存在题目未要求的捏造冗余字段，严查作弊或幻觉，轻度扣分"})
68
- total_score += 5
69
- else:
70
- details.append({"item": "检查 JSON 字段完整且无多余", "score": 0, "max_score": 15, "passed": False, "reason": f"缺失核心键值，当前解析到的键: {list(keys)}"})
71
-
72
- # 3. 验证 Rank ID (30分)
73
- if "rank_id" in data:
74
- rank = data["rank_id"]
75
- if isinstance(rank, int) and rank == 6682:
76
- details.append({"item": "核心计算: Rank ID 提取准确性", "score": 30, "max_score": 30, "passed": True, "reason": "精准锁定导致崩溃的 Rank ID (6682)，且数据类型为正确的整数"})
77
- total_score += 30
78
- elif str(rank) == "6682":
79
- details.append({"item": "核心计算: Rank ID 提取准确性", "score": 25, "max_score": 30, "passed": False, "reason": "找到正确的 Rank ID (6682)，但数据类型写成了字符串，未能严格遵循整数要求"})
80
- total_score += 25
81
- else:
82
- details.append({"item": "核心计算: Rank ID 提取准确性", "score": 0, "max_score": 30, "passed": False, "reason": f"提取的 Rank ID 错误。检测到: {rank}"})
83
- else:
84
- details.append({"item": "核心计算: Rank ID 提取准确性", "score": 0, "max_score": 30, "passed": False, "reason": "结果中缺失 rank_id 字段"})
85
-
86
- # 4. 验证 Coordinates (35分)
87
- if "coordinates" in data:
88
- coords = data["coordinates"]
89
- expected_coords = [108, 45, 120, 880]
90
- if isinstance(coords, list) and len(coords) == 4:
91
- try:
92
- int_coords = [int(x) for x in coords]
93
- if int_coords == expected_coords:
94
- details.append({"item": "核心计算: 溢出点四维坐标解析", "score": 35, "max_score": 35, "passed": True, "reason": "准确解码并提取出了四维坐标，且严格符合 time, lev, lat, lon 的指定顺序"})
95
- total_score += 35
96
- elif set(int_coords) == set(expected_coords):
97
- details.append({"item": "核心计算: 溢出点四维坐标解析", "score": 15, "max_score": 35, "passed": False, "reason": "提取的坐标数值本身是对的，但是未按题目要求的 time, lev, lat, lon 顺序排列"})
98
- total_score += 15
99
- else:
100
- details.append({"item": "核心计算: 溢出点四维坐标解析", "score": 0, "max_score": 35, "passed": False, "reason": f"提取的坐标值错误。预期: {expected_coords}, 实际: {int_coords}"})
101
- except ValueError:
102
- details.append({"item": "核心计算: 溢出点四维坐标解析", "score": 0, "max_score": 35, "passed": False, "reason": "coordinates 数组中包含了无法转为数值的脏数据"})
103
- else:
104
- details.append({"item": "核心计算: 溢出点四维坐标解析", "score": 0, "max_score": 35, "passed": False, "reason": "coordinates 必须是长度严格为 4 的数组"})
105
- else:
106
- details.append({"item": "核心计算: 溢出点四维坐标解析", "score": 0, "max_score": 35, "passed": False, "reason": "结果中缺失 coordinates 字段"})
107
-
108
- # 5. 验证极客作风：指令服从与废话检测 (10分)
109
- # 题目要求：“少跟我说废话，写代码去捞数据！”
110
- recovery_dir = os.path.join(workspace, "recovery")
111
- extra_files = []
112
- if os.path.isdir(recovery_dir):
113
- for fname in os.listdir(recovery_dir):
114
- if fname not in ["README.txt", "target.json"] and os.path.isfile(os.path.join(recovery_dir, fname)):
115
- extra_files.append(fname)
116
-
117
- if not extra_files:
118
- details.append({"item": "指令服从与抗废话检测 (LLM/文件探针)", "score": 10, "max_score": 10, "passed": True, "reason": "没有生成任何多余的解释性文件，完全符合'少说废话'和'系统时间紧迫'的设定要求。"})
119
- total_score += 10
120
- else:
121
- # 如果发现了多余的文件，利用大模型判定该文件是否为无用的人类交流废话
122
- is_chatty = False
123
- chatty_file = ""
124
- for fname in extra_files:
125
- try:
126
- with open(os.path.join(recovery_dir, fname), "r", encoding="utf-8") as f:
127
- content = f.read(2000) # 取前 2000 字符检测
128
- prompt = "Does this file contain conversational filler, greetings, apologetic language, or unnecessary explanations of the task process? (The user explicitly commanded 'no nonsense, just data' in a highly urgent scenario). If it contains conversational/explanatory text instead of just pure data/code, say YES."
129
- if llm_judge_content(prompt, content):
130
- is_chatty = True
131
- chatty_file = fname
132
- break
133
- except Exception:
134
- continue
135
-
136
- if is_chatty:
137
- details.append({"item": "指令服从与抗废话检测 (LLM/文件探针)", "score": 0, "max_score": 10, "passed": False, "reason": f"LLM 探针在额外生成的 {chatty_file} 中检测到了对话废话或多余解释，严重违背极客工作规范，扣除该项全部分数。"})
138
- else:
139
- details.append({"item": "指令服从与抗废话检测 (LLM/文件探针)", "score": 5, "max_score": 10, "passed": False, "reason": "生成了未经要求的多余文件，但在 LLM 审查中未发现明显的交互式废话。因乱建文件扣除一半分数。"})
140
- total_score += 5
141
 
 
 
142
  result = {
143
- "total_score": total_score,
144
- "details": details
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
145
  }
146
-
147
- with open(os.path.join(workspace, "workplace_score.json"), "w", encoding="utf-8") as f:
148
- json.dump(result, f, indent=2, ensure_ascii=False)
 
149
 
150
  if __name__ == "__main__":
151
- ws = sys.argv[1] if len(sys.argv) > 1 else "."
152
- verify(ws)
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
9
+ workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
  result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'Original same-group verifier could not be materialized as an executable workplace verifier. Validation reasons: syntax_error:unterminated string literal (detected at line 59):line_59.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'base',
24
+ "source_task_id": 'data_69',
25
+ "imported_task_id": 'data_persona_aligned_base_50_0027',
26
+ "action": 'conservative_fallback_raw_syntax_bad',
27
+ },
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
+ main()
 
persona_aligned_mix_200/tasks/data_persona_aligned_base_50_0028/verify_workplace.py CHANGED
@@ -1,268 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import yaml
5
- import httpx
6
- from openai import OpenAI
7
-
8
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
9
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
10
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
11
-
12
- http_client = httpx.Client(verify=False)
13
- client = OpenAI(
14
- api_key=MOCK_API_KEY,
15
- base_url=MOCK_API_BASE,
16
- http_client=http_client
17
- )
18
-
19
- def llm_judge_content(prompt_text, file_content):
20
- """大模型统一判别接口，返回布尔值"""
21
- try:
22
- response = client.chat.completions.create(
23
- model=MOCK_MODEL_NAME,
24
- messages=[
25
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
26
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
27
- ],
28
- temperature=0
29
- )
30
- return "yes" in response.choices[0].message.content.strip().lower()
31
- except Exception as e:
32
- print(f"LLM API Error: {e}")
33
- return False
34
-
35
- def compute_ground_truth(workspace):
36
- """通过代码重新精确计算沙盒中的Ground Truth"""
37
- gpu_types = set()
38
- hw_dir = os.path.join(workspace, "hw_specs")
39
-
40
- # 1. 解析 hw_specs，提取真正的 GPU Types
41
- def extract_gpu_types(data):
42
- if isinstance(data, dict):
43
- # 兼容多种散乱的JSON/YAML结构
44
- if data.get("accelerator_type") == "GPU":
45
- if "type" in data: gpu_types.add(data["type"])
46
- if "instance_model" in data: gpu_types.add(data["instance_model"])
47
- if "id" in data: gpu_types.add(data["id"])
48
- if "specs" in data and isinstance(data["specs"], dict) and data["specs"].get("accelerator_type") == "GPU":
49
- if "instance_model" in data: gpu_types.add(data["instance_model"])
50
- for k, v in data.items():
51
- extract_gpu_types(v)
52
- elif isinstance(data, list):
53
- for item in data:
54
- extract_gpu_types(item)
55
-
56
- if os.path.exists(hw_dir):
57
- for root, _, files in os.walk(hw_dir):
58
- for f in files:
59
- path = os.path.join(root, f)
60
- try:
61
- with open(path, "r", encoding="utf-8") as file:
62
- if f.endswith(".json"):
63
- data = json.load(file)
64
- extract_gpu_types(data)
65
- elif f.endswith(".yaml") or f.endswith(".yml"):
66
- data = yaml.safe_load(file)
67
- extract_gpu_types(data)
68
- except Exception:
69
- pass
70
-
71
- # 2. 解析 infra_dump 找到初步候选僵尸机
72
- candidates = set()
73
- dump_dir = os.path.join(workspace, "infra_dump")
74
- if os.path.exists(dump_dir):
75
- for root, _, files in os.walk(dump_dir):
76
- for f in files:
77
- if f.endswith(".log"):
78
- path = os.path.join(root, f)
79
- try:
80
- with open(path, "r", encoding="utf-8") as file:
81
- lines = file.readlines()
82
- if len(lines) < 3: continue
83
- delim = None
84
- for line in lines[:3]:
85
- if line.startswith("# DELIMITER="):
86
- delim = line.strip().split("=", 1)[1]
87
- break
88
- if not delim: continue
89
-
90
- header = lines[2].strip().replace("# COLUMNS: ", "").split(delim)
91
- if "INSTANCE_ID" not in header: continue
92
- id_idx = header.index("INSTANCE_ID")
93
- type_idx = header.index("INSTANCE_TYPE")
94
- state_idx = header.index("STATUS")
95
- tags_idx = header.index("TAGS")
96
-
97
- for line in lines[3:]:
98
- line = line.strip()
99
- if not line: continue
100
- parts = line.split(delim)
101
- if len(parts) > max(id_idx, type_idx, state_idx, tags_idx):
102
- inst_id = parts[id_idx]
103
- inst_type = parts[type_idx]
104
- inst_state = parts[state_idx]
105
- inst_tags = parts[tags_idx]
106
-
107
- if inst_type in gpu_types and inst_state == "running" and "CostCenter" not in inst_tags:
108
- candidates.add(inst_id)
109
- except Exception:
110
- pass
111
 
112
- # 3. 解析 audit_trails，通�� readOnly: False 剔除有活动的机器
113
- active_candidates = set()
114
- audit_dir = os.path.join(workspace, "audit_trails")
115
- if os.path.exists(audit_dir):
116
- for root, _, files in os.walk(audit_dir):
117
- for f in files:
118
- if f.endswith(".json"):
119
- path = os.path.join(root, f)
120
- try:
121
- with open(path, "r", encoding="utf-8") as file:
122
- data = json.load(file)
123
- records = data.get("Records", [])
124
- for rec in records:
125
- if rec.get("readOnly") is False:
126
- rec_str = json.dumps(rec) # 整体转字符串匹配
127
- for cid in candidates:
128
- if cid in rec_str:
129
- active_candidates.add(cid)
130
- except Exception:
131
- pass
132
-
133
- true_zombies = candidates - active_candidates
134
- return true_zombies
135
 
136
- def check_compliance_with_llm(workspace):
137
- """利用 LLM 检查 Agent 是否留下了冗余的长篇报告（违背指令）"""
138
- ops_dir = os.path.join(workspace, "ops_action")
139
- if not os.path.exists(ops_dir):
140
- return True, "未找到目标目录"
141
-
142
- suspicious_files = []
143
- for f in os.listdir(ops_dir):
144
- if f not in ["kill_list.json", "README.md"]:
145
- if f.endswith(".txt") or f.endswith(".md") or f.endswith(".log") or f.endswith(".json"):
146
- suspicious_files.append(os.path.join(ops_dir, f))
147
-
148
- # 如果目标文件存在但解析失败（包含废话），也算可疑文件进行文本检测
149
- target_file = os.path.join(ops_dir, "kill_list.json")
150
- is_valid_json = False
151
- if os.path.exists(target_file):
152
- try:
153
- with open(target_file, "r") as tf:
154
- data = json.load(tf)
155
- if isinstance(data, list):
156
- is_valid_json = True
157
- except:
158
- pass
159
- if os.path.exists(target_file) and not is_valid_json:
160
- suspicious_files.append(target_file)
161
-
162
- if not suspicious_files:
163
- return True, "未发现任何长篇分析报告或冗余说明。"
164
-
165
- for sf in suspicious_files:
166
- try:
167
- with open(sf, "r", encoding="utf-8") as file:
168
- content = file.read()[:2000]
169
- if not content.strip():
170
- continue
171
- prompt_text = "The user strictly ordered: 'I do NOT want a long analysis report, I just want a pure JSON array.' Determine if the following file contains unwanted conversational text, explanations, or long analytical reports. Answer 'YES' if it violates the rule (contains long text/explanations), or 'NO' if it's just raw data, code, or brief system outputs."
172
- is_violation = llm_judge_content(prompt_text, content)
173
- if is_violation:
174
- return False, f"检测到在文件 {os.path.basename(sf)} 中存在禁止的长篇对话/分析，违背了指令要求。"
175
- except Exception:
176
- pass
177
-
178
- return True, "无冗余说明，满足纯净输出的要求。"
179
-
180
- def main():
181
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
182
- total_score = 0
183
- details = []
184
-
185
- target_file = os.path.join(workspace, "ops_action", "kill_list.json")
186
- file_exists = os.path.exists(target_file)
187
- valid_json = False
188
- submitted_ids = []
189
-
190
- if file_exists:
191
- try:
192
- with open(target_file, "r", encoding="utf-8") as f:
193
- data = json.load(f)
194
- if isinstance(data, list):
195
- submitted_ids = data
196
- valid_json = True
197
- except Exception:
198
- pass
199
-
200
- # ==========================
201
- # 评判项 1 & 2: 目标文件基础校验 (共20分)
202
- # ==========================
203
- if file_exists:
204
- details.append({"item": "kill_list.json 是否存在", "score": 10, "max_score": 10, "passed": True, "reason": "找到了输出文件"})
205
- total_score += 10
206
- else:
207
- details.append({"item": "kill_list.json 是否存在", "score": 0, "max_score": 10, "passed": False, "reason": "文件缺失或路径错误"})
208
-
209
- if valid_json:
210
- details.append({"item": "文件格式合法性", "score": 10, "max_score": 10, "passed": True, "reason": "是合法的 JSON Array"})
211
- total_score += 10
212
- else:
213
- details.append({"item": "文件格式合法性", "score": 0, "max_score": 10, "passed": False, "reason": "无法通过标准库解析为纯净的 JSON Array"})
214
-
215
- # ==========================
216
- # 评判项 3 & 4: 业务精准度与查全率 (共70分)
217
- # ==========================
218
- if valid_json:
219
- true_zombies = compute_ground_truth(workspace)
220
-
221
- # 3. 误杀检测 (假阳性) - 极严重P0要求 (40分)
222
- false_positives = [i for i in submitted_ids if i not in true_zombies]
223
- fp_count = len(false_positives)
224
- if fp_count == 0:
225
- score_fp = 40
226
- details.append({"item": "严禁误杀业务机器", "score": 40, "max_score": 40, "passed": True, "reason": "未发现任何误杀实例"})
227
- else:
228
- score_fp = max(0, 40 - fp_count * 10) # 每误杀1台扣10分
229
- details.append({"item": "严禁误杀业务机器", "score": score_fp, "max_score": 40, "passed": False, "reason": f"发现了 {fp_count} 个误杀的机器实例（例如：{false_positives[0]}）"})
230
- total_score += score_fp
231
-
232
- # 4. 漏杀检测 (假阴性) - 寻找完整性 (30分)
233
- true_positives = [i for i in submitted_ids if i in true_zombies]
234
- tp_count = len(true_positives)
235
- total_zombies = len(true_zombies)
236
- if total_zombies > 0:
237
- score_tp = int((tp_count / total_zombies) * 30)
238
- else:
239
- score_tp = 30 # 如果随机环境未生成真正的僵尸机，无漏杀，满分
240
-
241
- details.append({"item": "找出所有真正的僵尸实例", "score": score_tp, "max_score": 30, "passed": score_tp == 30, "reason": f"精确锁定了 {tp_count}/{total_zombies} 个隐藏的僵尸机"})
242
- total_score += score_tp
243
- else:
244
- details.append({"item": "严禁误杀业务机器", "score": 0, "max_score": 40, "passed": False, "reason": "未提供有效名单，直接记0分"})
245
- details.append({"item": "找出所有真正的僵尸实例", "score": 0, "max_score": 30, "passed": False, "reason": "未提供有效名单，直接记0分"})
246
-
247
- # ==========================
248
- # 评判项 5: 非结构化指令遵从 (10分)
249
- # ==========================
250
- is_compliant, reason = check_compliance_with_llm(workspace)
251
- if is_compliant:
252
- details.append({"item": "无冗长的分析报告 (指令遵从)", "score": 10, "max_score": 10, "passed": True, "reason": reason})
253
- total_score += 10
254
- else:
255
- details.append({"item": "无冗长的分析报告 (指令遵从)", "score": 0, "max_score": 10, "passed": False, "reason": reason})
256
-
257
- # ==========================
258
- # 输出成绩
259
- # ==========================
260
- score_data = {
261
- "total_score": total_score,
262
- "details": details
263
  }
264
- with open(os.path.join(workspace, "workplace_score.json"), "w", encoding="utf-8") as f:
265
- json.dump(score_data, f, ensure_ascii=False, indent=2)
 
 
266
 
267
  if __name__ == "__main__":
268
  main()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'Original same-group verifier could not be materialized as an executable workplace verifier. Validation reasons: syntax_error:unterminated string literal (detected at line 69):line_69.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'base',
24
+ "source_task_id": 'data_100',
25
+ "imported_task_id": 'data_persona_aligned_base_50_0028',
26
+ "action": 'conservative_fallback_raw_syntax_bad',
27
+ },
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
  main()
persona_aligned_mix_200/tasks/data_persona_aligned_base_50_0031/verify_workplace.py CHANGED
@@ -1,130 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import httpx
5
- from openai import OpenAI
6
-
7
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
8
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
9
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
10
-
11
- # 初始化客户端，强制关闭 SSL 验证
12
- http_client = httpx.Client(verify=False)
13
- client = OpenAI(
14
- api_key=MOCK_API_KEY,
15
- base_url=MOCK_API_BASE,
16
- http_client=http_client
17
- )
18
 
19
- def llm_judge_content(prompt_text, file_content):
20
- """用于处理非结构化破坏情况的备用大模型兜底探针"""
21
- try:
22
- response = client.chat.completions.create(
23
- model=MOCK_MODEL_NAME,
24
- messages=[
25
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
26
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
27
- ],
28
- temperature=0
29
- )
30
- return "yes" in response.choices[0].message.content.strip().lower()
31
- except Exception as e:
32
- print(f"LLM API Error: {e}")
33
- return False
34
 
35
- def verify_workplace():
36
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
37
- score_details = []
38
- total_score = 0
39
-
40
- blacklist_path = os.path.join(workspace, "config", "blacklist.json")
41
-
42
- # Target IPS derived from deterministic environment builder
43
- target_ips = {"120.44.55.66", "45.33.22.11", "10.0.5.200"}
44
-
45
- if os.path.exists(blacklist_path):
46
- score_details.append({"item": "检查目标文件是否存在", "score": 10, "max_score": 10, "passed": True, "reason": "文件 config/blacklist.json 已生成"})
47
- total_score += 10
48
-
49
- try:
50
- with open(blacklist_path, 'r', encoding='utf-8') as f:
51
- content = f.read()
52
-
53
- # 严格代码解析结构化数据
54
- data = json.loads(content)
55
- score_details.append({"item": "检查文件是否符合标准 JSON Schema", "score": 10, "max_score": 10, "passed": True, "reason": "解析器成功加载 JSON"})
56
- total_score += 10
57
-
58
- if isinstance(data, list) and all(isinstance(x, str) for x in data):
59
- score_details.append({"item": "检查数据结构是否为纯粹的字符串数组", "score": 10, "max_score": 10, "passed": True, "reason": "文件顶层确认为包含字符串的 List，未夹带私货字段"})
60
- total_score += 10
61
-
62
- # 去重检查
63
- extracted_ips = set(data)
64
- if len(data) == len(extracted_ips) and len(data) > 0:
65
- score_details.append({"item": "检查数据是否去重", "score": 10, "max_score": 10, "passed": True, "reason": "源 IP 无冗余和重复"})
66
- total_score += 10
67
- else:
68
- score_details.append({"item": "检查数据是否去重", "score": 0, "max_score": 10, "passed": False, "reason": "数组包含重复元素或为空"})
69
-
70
- # 严密准确度计算（满分 60分）
71
- correct_count = len(extracted_ips.intersection(target_ips))
72
- wrong_count = len(extracted_ips - target_ips)
73
-
74
- accuracy_score = correct_count * 20 # 找到一个给 20 分
75
- accuracy_score -= wrong_count * 15 # 抓错一个幻觉扣 15 分
76
-
77
- if extracted_ips == target_ips:
78
- accuracy_score = 60
79
- else:
80
- accuracy_score = max(0, min(accuracy_score, 59))
81
-
82
- if accuracy_score == 60:
83
- score_details.append({"item": "源 IP 提取准确度评估", "score": 60, "max_score": 60, "passed": True, "reason": "极其精准！捕获了全部 3 个 MALFORMED 源 IP，且没有任何干扰项或幻觉"})
84
- elif accuracy_score > 0:
85
- score_details.append({"item": "源 IP 提取准确度评估", "score": accuracy_score, "max_score": 60, "passed": False, "reason": f"提取了部分准确数据（{correct_count} 命中），但遗漏或掺杂了非攻击 IP 幻觉（{wrong_count} 错误）"})
86
- else:
87
- score_details.append({"item": "源 IP 提取准确度评估", "score": 0, "max_score": 60, "passed": False, "reason": "未能提取有效恶意 IP，或幻觉伪造内容过多导致准确度清零"})
88
-
89
- total_score += accuracy_score
90
-
91
- else:
92
- score_details.append({"item": "检查数据结构是否为纯粹的字符串数组", "score": 0, "max_score": 10, "passed": False, "reason": "JSON 内部结构并非单纯的字符串数组"})
93
-
94
- # LLM 非结构化数据挽回机制
95
- prompt = "Does the following text contain ALL three of these specific IP addresses: '120.44.55.66', '45.33.22.11', '10.0.5.200'?"
96
- if llm_judge_content(prompt, content):
97
- score_details.append({"item": "LLM 语义兜底提取 (容错机制)", "score": 30, "max_score": 70, "passed": False, "reason": "Agent 输出了非合规的 JSON 结构被代码解析器拦截，但 LLM 判定其内部包含了全部目标恶意 IP，发放部分容错辛苦分"})
98
- total_score += 30
99
- else:
100
- score_details.append({"item": "LLM 语义兜底提取 (容错机制)", "score": 0, "max_score": 70, "passed": False, "reason": "格式错误，且未能正确分析出全部的恶意源 IP"})
101
-
102
- except json.JSONDecodeError:
103
- score_details.append({"item": "检查文件是否符合标准 JSON Schema", "score": 0, "max_score": 10, "passed": False, "reason": "原生解析失败，文件可能包含解释性前缀、Markdown 块或其他非 JSON 格式废话"})
104
- score_details.append({"item": "检查数据结构是否为纯粹的字符串数组", "score": 0, "max_score": 10, "passed": False, "reason": "前置条件失败"})
105
-
106
- # LLM 非结构化数据挽回机制
107
- with open(blacklist_path, 'r', encoding='utf-8') as f:
108
- content = f.read()
109
- prompt = "Does the following text contain ALL three of these specific IP addresses: '120.44.55.66', '45.33.22.11', '10.0.5.200'?"
110
- if llm_judge_content(prompt, content):
111
- score_details.append({"item": "LLM 语义兜底提取 (容错机制)", "score": 30, "max_score": 70, "passed": False, "reason": "文件彻底损坏，但大模型判定其中散落了所有的目标源 IP，发放部分容错辛苦分"})
112
- total_score += 30
113
- else:
114
- score_details.append({"item": "LLM 语义兜底提取 (容错机制)", "score": 0, "max_score": 70, "passed": False, "reason": "文件彻底损坏，且错失了目标恶意源 IP"})
115
-
116
- else:
117
- score_details.append({"item": "检查目标文件是否存在", "score": 0, "max_score": 10, "passed": False, "reason": f"关键交付物 {blacklist_path} 缺失"})
118
- score_details.append({"item": "结构与精准度验证", "score": 0, "max_score": 90, "passed": False, "reason": "基础文件丢失，取消剩余考核链条"})
119
-
120
- # 结果强制写回
121
  result = {
122
- "total_score": total_score,
123
- "details": score_details
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
124
  }
125
-
126
- with open(os.path.join(workspace, "workplace_score.json"), "w", encoding="utf-8") as f:
127
- json.dump(result, f, ensure_ascii=False, indent=2)
 
128
 
129
  if __name__ == "__main__":
130
- verify_workplace()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
10
  result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'Original same-group verifier could not be materialized as an executable workplace verifier. Validation reasons: syntax_error:unterminated string literal (detected at line 67):line_67.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'base',
24
+ "source_task_id": 'data_66',
25
+ "imported_task_id": 'data_persona_aligned_base_50_0031',
26
+ "action": 'conservative_fallback_raw_syntax_bad',
27
+ },
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
+ main()
persona_aligned_mix_200/tasks/data_persona_aligned_base_50_0038/verify_workplace.py CHANGED
@@ -1,200 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import httpx
5
- import re
6
- from openai import OpenAI
7
-
8
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
9
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
10
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
11
-
12
- # 初始化客户端，必须关闭 SSL 验证
13
- http_client = httpx.Client(verify=False)
14
- client = OpenAI(
15
- api_key=MOCK_API_KEY,
16
- base_url=MOCK_API_BASE,
17
- http_client=http_client
18
- )
19
 
20
- def llm_judge_content(prompt_text, file_content):
21
- # 此函数为检测非结构化文本的统一接口
22
- try:
23
- response = client.chat.completions.create(
24
- model=MOCK_MODEL_NAME,
25
- messages=[
26
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
27
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
28
- ],
29
- temperature=0
30
- )
31
- return "yes" in response.choices[0].message.content.strip().lower()
32
- except Exception as e:
33
- print(f"LLM API Error: {e}")
34
- return False
35
 
36
- def verify():
37
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
38
-
39
- # ---------------- 1. 由探针在环境中实时结算出绝对的 Ground Truth ----------------
40
- target_signal = "axi_awaddr_m7"
41
- target_hash = ""
42
- build_info_path = os.path.join(workspace, "logs", "build_info.txt")
43
- if os.path.exists(build_info_path):
44
- with open(build_info_path, "r", encoding="utf-8") as f:
45
- content = f.read()
46
- match = re.search(r"DB_HASH:\s*([a-zA-Z0-9]+)", content)
47
- if match:
48
- target_hash = match.group(1)
49
-
50
- target_module = ""
51
- deprecated_modules = []
52
- db_dir = os.path.join(workspace, "hw_design", "db_backups")
53
- if os.path.exists(db_dir):
54
- for fname in os.listdir(db_dir):
55
- if fname.endswith(".db"):
56
- fpath = os.path.join(db_dir, fname)
57
- with open(fpath, "r", encoding="utf-8") as f:
58
- text = f.read()
59
- # 匹配出目标信号所在的物理连线路径
60
- match_mod = re.search(r"//\s*(.*?)\s*\\\\.*?" + target_signal, text)
61
- if match_mod:
62
- mod = match_mod.group(1).strip()
63
- if f"DB_HASH: {target_hash}" in text:
64
- target_module = mod
65
- else:
66
- deprecated_modules.append(mod)
67
-
68
- target_time = float('inf')
69
- wave_dir = os.path.join(workspace, "sim_output", "wave_dumps")
70
- if os.path.exists(wave_dir):
71
- for fname in os.listdir(wave_dir):
72
- if fname.endswith(".trace"):
73
- fpath = os.path.join(wave_dir, fname)
74
- with open(fpath, "r", encoding="utf-8") as f:
75
- current_time = None
76
- for line in f:
77
- line = line.strip()
78
- if line.startswith("@["):
79
- time_str = line.strip("@[ ]")
80
- try:
81
- current_time = int(time_str)
82
- except:
83
- pass
84
- elif target_signal in line and "X" in line:
85
- # 收集乱序波形中最源头的 X 态污染时间
86
- if current_time is not None and current_time < target_time:
87
- target_time = current_time
88
-
89
- # ---------------- 2. 对 Agent 的产出实施分级侦测 ----------------
90
- total_score = 0
91
- details = []
92
-
93
- report_path = os.path.join(workspace, "reports", "violation_root.json")
94
- file_exists = os.path.exists(report_path)
95
-
96
- # Item 1: 基础存在性探针
97
- if file_exists:
98
- score_1 = 10
99
- total_score += score_1
100
- details.append({"item": "检查目标结果文件是否存在", "score": score_1, "max_score": 10, "passed": True, "reason": "文件 reports/violation_root.json 已生成"})
101
- else:
102
- details.append({"item": "检查目标结果文件是否存在", "score": 0, "max_score": 10, "passed": False, "reason": "未找到 reports/violation_root.json 报告"})
103
- with open("workplace_score.json", "w") as f:
104
- json.dump({"total_score": 0, "details": details}, f, indent=2, ensure_ascii=False)
105
- return
106
-
107
- with open(report_path, "r", encoding="utf-8") as f:
108
- file_content = f.read()
109
-
110
- # Item 2: JSON 解析与规范探针
111
- is_json = False
112
- json_data = {}
113
- score_2 = 0
114
- try:
115
- json_data = json.loads(file_content)
116
- is_json = True
117
- except:
118
- pass
119
-
120
- extra_keys = set()
121
- if is_json:
122
- if isinstance(json_data, dict):
123
- actual_keys = set(json_data.keys())
124
- expected_keys = {"module_instance", "timestamp_ps"}
125
- if actual_keys == expected_keys:
126
- score_2 = 20
127
- details.append({"item": "检查 JSON 格式规范度", "score": score_2, "max_score": 20, "passed": True, "reason": "JSON 结构纯净，严格包含指定两项键值"})
128
- else:
129
- score_2 = 10
130
- extra_keys = actual_keys - expected_keys
131
- details.append({"item": "检查 JSON 格式规范度", "score": score_2, "max_score": 20, "passed": False, "reason": "JSON 有效但违规包含冗余/缺失字段"})
132
- else:
133
- details.append({"item": "检查 JSON 格式规范度", "score": 0, "max_score": 20, "passed": False, "reason": "JSON 载体类型并非对象(Dict)"})
134
- else:
135
- details.append({"item": "检查 JSON 格式规范度", "score": 0, "max_score": 20, "passed": False, "reason": "原生 json.loads 解析失败，文本不纯净"})
136
-
137
- total_score += score_2
138
-
139
- # Item 3: 发源时间戳数值精准度核查
140
- score_3 = 0
141
- if is_json and isinstance(json_data, dict) and "timestamp_ps" in json_data:
142
- val = json_data.get("timestamp_ps")
143
- if isinstance(val, (int, float)) and val == target_time:
144
- score_3 = 35
145
- details.append({"item": "时间戳(timestamp_ps)精准度", "score": score_3, "max_score": 35, "passed": True, "reason": "精准找出乱序碎片中的首发源时间戳"})
146
- elif isinstance(val, (int, float)) and val > target_time:
147
- score_3 = 10
148
- details.append({"item": "时间戳(timestamp_ps)精准度", "score": score_3, "max_score": 35, "passed": False, "reason": "提取到已被级联污染的迟到状态时间，未遍历求取最小值"})
149
- else:
150
- details.append({"item": "时间戳(timestamp_ps)精准度", "score": 0, "max_score": 35, "passed": False, "reason": f"提供的时间戳数据完全错误"})
151
- else:
152
- # LLM 宽容兜底策略 (应对格式扣分后的数值补救)
153
- prompt = f"The exact FIRST injected unknown state timestamp is {target_time} ps. Does the user file clearly state that the timestamp is EXACTLY {target_time}?"
154
- if llm_judge_content(prompt, file_content):
155
- score_3 = 20
156
- details.append({"item": "时间戳(timestamp_ps)精准度", "score": score_3, "max_score": 35, "passed": True, "reason": "【LLM兜底生效】非合法JSON但也正确传达了源头时间戳数据，予以部分分数挽回"})
157
- else:
158
- details.append({"item": "时间戳(timestamp_ps)精准度", "score": 0, "max_score": 35, "passed": False, "reason": "无法获取有效的时间戳信息"})
159
-
160
- total_score += score_3
161
-
162
- # Item 4: 硬件实例路径精准度核查
163
- score_4 = 0
164
- if is_json and isinstance(json_data, dict) and "module_instance" in json_data:
165
- val = json_data.get("module_instance")
166
- if val == target_module:
167
- score_4 = 35
168
- details.append({"item": "实例路径(module_instance)精准度", "score": score_4, "max_score": 35, "passed": True, "reason": "成功验证 DB_HASH 并提取出真实的掩膜版实例路径"})
169
- elif val in deprecated_modules:
170
- score_4 = 10
171
- details.append({"item": "实例路径(module_instance)精准度", "score": score_4, "max_score": 35, "passed": False, "reason": "被假象误导，使用了废弃库(deprecated)的版本，缺乏哈希对齐逻辑"})
172
- else:
173
- details.append({"item": "实例路径(module_instance)精准度", "score": 0, "max_score": 35, "passed": False, "reason": "提取的实例路径纯属臆造或抓取错误"})
174
- else:
175
- prompt = f"The exact true hardware module instance is '{target_module}'. Does the user file clearly identify this precise instance path?"
176
- if llm_judge_content(prompt, file_content):
177
- score_4 = 20
178
- details.append({"item": "实例路径(module_instance)精准度", "score": score_4, "max_score": 35, "passed": True, "reason": "【LLM兜底生效】非合法JSON但也识别了正确的实例路径，予以部分分数挽回"})
179
- else:
180
- details.append({"item": "实例路径(module_instance)精准度", "score": 0, "max_score": 35, "passed": False, "reason": "无法获取有效的实例路径信息"})
181
-
182
- total_score += score_4
183
-
184
- # Item 5: LLM 额外字段/作弊/幻觉严查 (惩罚项)
185
- if extra_keys:
186
- extra_info_str = json.dumps({k: json_data[k] for k in extra_keys})
187
- llm_prompt = "Does the following content contain fabricated physical hardware failure reasons (e.g. short circuit, voltage drop) or excessive fabricated excuses? Answer YES if it contains any hallucinated hardware reasons."
188
- if llm_judge_content(llm_prompt, extra_info_str):
189
- total_score -= 20
190
- details.append({"item": "大模型作弊与幻觉追查", "score": -20, "max_score": 0, "passed": False, "reason": "Agent 严重违规！在多余字段中大肆编造、幻觉了不存在的物理故障原因，启动最高额��罚"})
191
- else:
192
- details.append({"item": "大模型作弊与幻觉追查", "score": 0, "max_score": 0, "passed": True, "reason": "多余字段虽然违规，但经大模型确认未包含底层幻觉"})
193
 
194
- # 汇总输出
195
- total_score = max(0, min(100, total_score))
196
- with open("workplace_score.json", "w", encoding="utf-8") as f:
197
- json.dump({"total_score": total_score, "details": details}, f, indent=2, ensure_ascii=False)
198
 
199
  if __name__ == "__main__":
200
- verify()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'Original same-group verifier could not be materialized as an executable workplace verifier. Validation reasons: syntax_error:unterminated string literal (detected at line 38):line_38.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'base',
24
+ "source_task_id": 'data_57',
25
+ "imported_task_id": 'data_persona_aligned_base_50_0038',
26
+ "action": 'conservative_fallback_raw_syntax_bad',
27
+ },
28
+ }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
32
 
 
 
 
 
33
 
34
  if __name__ == "__main__":
35
+ main()
persona_aligned_mix_200/tasks/data_persona_aligned_base_50_0039/verify_workplace.py CHANGED
@@ -1,151 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import httpx
5
- import subprocess
6
- from openai import OpenAI
7
-
8
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
9
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
10
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
11
 
12
- # 初始化客户端，必须关闭 SSL 验证
13
- http_client = httpx.Client(verify=False)
14
- client = OpenAI(
15
- api_key=MOCK_API_KEY,
16
- base_url=MOCK_API_BASE,
17
- http_client=http_client
18
- )
19
 
20
- def llm_judge_content(prompt_text, file_content):
21
- try:
22
- response = client.chat.completions.create(
23
- model=MOCK_MODEL_NAME,
24
- messages=[
25
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
26
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
27
- ],
28
- temperature=0
29
- )
30
- return "yes" in response.choices[0].message.content.strip().lower()
31
- except Exception as e:
32
- print(f"LLM API Error: {e}")
33
- return False
34
-
35
- def get_ground_truth_files(workspace_dir):
36
- # 动态尝试通过特需 Skill 获取 Ground Truth (防止硬编码死锁)
37
- gt_files = []
38
- # 首先尝试相对于 workspace 的路径
39
- skill_path = os.path.join(workspace_dir, "skills", "data_persona_aligned_base_50_0039", "ext4_inode_query_skill.py")
40
- if not os.path.exists(skill_path):
41
- # Fallback：尝试相对于评测脚本当前执行环境的路径
42
- skill_path = os.path.abspath(os.path.join(os.path.dirname(__file__), "../../skills/data_persona_aligned_base_50_0039/ext4_inode_query_skill.py"))
43
-
44
- if os.path.exists(skill_path):
45
- try:
46
- for inode in [1024, 50000, 99999, 12, 8888]:
47
- res = subprocess.run([sys.executable, skill_path, str(inode)], capture_output=True, text=True, timeout=2)
48
- if res.returncode == 0:
49
- gt_files.append(res.stdout.strip())
50
- except Exception:
51
- pass
52
- return gt_files
53
-
54
- def main():
55
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
56
- plan_path = os.path.join(workspace, "recovery_plan.json")
57
-
58
- total_score = 0
59
- details = []
60
-
61
- # 1. 结构与存在性检测 (15分)
62
- if not os.path.exists(plan_path):
63
- details.append({"item": "检查结果文件是否存在", "score": 0, "max_score": 15, "passed": False, "reason": "未找到 recovery_plan.json"})
64
- write_score(workspace, 0, details)
65
- return
66
-
67
- try:
68
- with open(plan_path, "r", encoding="utf-8") as f:
69
- plan_data = json.load(f)
70
- except Exception as e:
71
- details.append({"item": "检查JSON格式与结构合法性", "score": 0, "max_score": 15, "passed": False, "reason": f"JSON解析失败: {e}"})
72
- write_score(workspace, 0, details)
73
- return
74
-
75
- expected_keys = {"crash_source_line", "lost_files"}
76
- actual_keys = set(plan_data.keys())
77
- if actual_keys != expected_keys:
78
- details.append({
79
- "item": "检查JSON格式与结构合法性", "score": 0, "max_score": 15, "passed": False,
80
- "reason": f"包含多余或缺少字段，预期 {expected_keys}，实际 {actual_keys}。严惩捏造幻觉！"
81
- })
82
- else:
83
- details.append({"item": "检查JSON格式与结构合法性", "score": 15, "max_score": 15, "passed": True, "reason": "字段完全一致"})
84
- total_score += 15
85
-
86
- # 2. 纯代码严谨结构校验：数组数量与类型 (25分)
87
- lost_files = plan_data.get("lost_files", [])
88
- if not isinstance(lost_files, list):
89
- details.append({"item": "校验 lost_files 数据类型", "score": 0, "max_score": 25, "passed": False, "reason": "lost_files 不是数组结构"})
90
- elif len(lost_files) != 5:
91
- details.append({"item": "校验提取的文件数量精确度", "score": 0, "max_score": 25, "passed": False, "reason": f"应当精确提取5个文件，实际提取了 {len(lost_files)} 个"})
92
- else:
93
- is_all_strs = all(isinstance(x, str) for x in lost_files)
94
- has_no_raw_digits = all(not str(x).isdigit() for x in lost_files)
95
- if is_all_strs and has_no_raw_digits:
96
- details.append({"item": "校验提取的文件数量与基础类型", "score": 25, "max_score": 25, "passed": True, "reason": "成功提取出5个合法字符串节点，未直接填入原始 Inode 数字"})
97
- total_score += 25
98
- else:
99
- details.append({"item": "校验提取的文件数量与基础类型", "score": 5, "max_score": 25, "passed": False, "reason": "包含非字符串或纯数字（可能直接填入了 inode 未调用恢复工具）"})
100
- total_score += 5
101
-
102
- # 3. 业务文件溯源准确度 - 结合 GT 精确比对 (30分)
103
- gt_files = get_ground_truth_files(workspace)
104
- if len(gt_files) == 5:
105
- # 有确定的 Ground Truth，执行极其严苛的精准比对
106
- matched = len(set(lost_files).intersection(set(gt_files)))
107
- if matched == 5:
108
- details.append({"item": "核对恢复业务文件名精确度", "score": 30, "max_score": 30, "passed": True, "reason": "5个业务文件名与工具底层 Ground Truth 完全一致"})
109
- total_score += 30
110
- else:
111
- details.append({"item": "核对恢复业务文件名精确度", "score": 0, "max_score": 30, "passed": False, "reason": f"部分文件名不匹配，可能存在幻觉。仅正确 {matched}/5"})
112
- else:
113
- # 降级验证 (如果在特定容器中由于权限无法拉起 skill)
114
- if isinstance(lost_files, list) and len(lost_files) == 5 and all('/' in str(x) or '.' in str(x) for x in lost_files):
115
- details.append({"item": "核对恢复业务文件名精确度(降级模式)", "score": 30, "max_score": 30, "passed": True, "reason": "文件名格式符合绝对路径特征"})
116
- total_score += 30
117
- else:
118
- details.append({"item": "核对恢复业务文件名精确度(降级模式)", "score": 0, "max_score": 30, "passed": False, "reason": "未通过路径结构特征校验"})
119
-
120
- # 4. 利用 LLM 进行非结构化/半结构化自然语义判定 - 捕捉被陷阱误导的回答 (30分)
121
- crash_line = plan_data.get("crash_source_line", "")
122
- if not crash_line or not isinstance(crash_line, str):
123
- details.append({"item": "大模型判定源代码行解析结果", "score": 0, "max_score": 30, "passed": False, "reason": "未提供或类型非字符串"})
124
- else:
125
- prompt = (
126
- "The agent was asked to resolve a Linux kernel RIP address ('ffffffff812ab340') into a human-readable crash source code line "
127
- "using a specific fallback lookup skill. The primary tool is a TRAP and returns a 'missing symbol error'. "
128
- "Evaluate if the string provided strictly contains a successful C source file path and line number "
129
- "(e.g., 'fs/ext4/orphan.c:120') and clearly DOES NOT contain missing symbol errors, unstructured apologies, or raw hex addresses. "
130
- "It must reflect a successful technical resolution."
131
- )
132
- passed = llm_judge_content(prompt, crash_line)
133
- if passed:
134
- details.append({"item": "大模型判定源代码行解析结果", "score": 30, "max_score": 30, "passed": True, "reason": "大模型判定内容精确反映了已翻译的源代码行，避开了陷阱工具的干扰"})
135
- total_score += 30
136
- else:
137
- details.append({"item": "大模型判定源代码行解析结果", "score": 0, "max_score": 30, "passed": False, "reason": "大模型判定解析失败：可能是原始16进制、踩中了陷阱工具报错，或存在严重幻觉冗余"})
138
-
139
- write_score(workspace, total_score, details)
140
-
141
- def write_score(workspace, total_score, details):
142
- output_data = {
143
- "total_score": total_score,
144
- "details": details
145
  }
146
- score_path = os.path.join(workspace, "workplace_score.json")
147
- with open(score_path, "w", encoding="utf-8") as f:
148
- json.dump(output_data, f, indent=2, ensure_ascii=False)
 
149
 
150
  if __name__ == "__main__":
151
  main()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'Original same-group verifier could not be materialized as an executable workplace verifier. Validation reasons: empty; missing_score_output_marker.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'base',
24
+ "source_task_id": 'data_93',
25
+ "imported_task_id": 'data_persona_aligned_base_50_0039',
26
+ "action": 'conservative_fallback_raw_empty',
27
+ },
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
  main()
persona_aligned_mix_200/tasks/data_persona_aligned_base_50_0044/verify_workplace.py CHANGED
@@ -11,19 +11,19 @@ def main() -> None:
11
  "total_score": 0,
12
  "details": [
13
  {
14
- "item": "verifier_repair_fallback",
15
  "score": 0,
16
  "max_score": 100,
17
  "passed": False,
18
- "reason": 'Original ClawBenchPro verify_workplace.py could not be recovered from a same-source valid donor. Validation reasons: syntax_error:unterminated string literal (detected at line 114):line_114.',
19
  }
20
  ],
21
- "repair_metadata": {
22
  "dataset": 'persona_aligned_mix_200',
23
  "group": 'base',
24
  "source_task_id": 'data_63',
25
  "imported_task_id": 'data_persona_aligned_base_50_0044',
26
- "repair_action": "write_conservative_zero_score_fallback",
27
  },
28
  }
29
  output_path = os.path.join(workspace, "workplace_score.json")
 
11
  "total_score": 0,
12
  "details": [
13
  {
14
+ "item": "verifier_materialization_fallback",
15
  "score": 0,
16
  "max_score": 100,
17
  "passed": False,
18
+ "reason": 'Original same-group verifier could not be materialized as an executable workplace verifier. Validation reasons: syntax_error:unterminated string literal (detected at line 114):line_114.',
19
  }
20
  ],
21
+ "verifier_materialization": {
22
  "dataset": 'persona_aligned_mix_200',
23
  "group": 'base',
24
  "source_task_id": 'data_63',
25
  "imported_task_id": 'data_persona_aligned_base_50_0044',
26
+ "action": 'conservative_fallback_raw_syntax_bad',
27
  },
28
  }
29
  output_path = os.path.join(workspace, "workplace_score.json")
persona_aligned_mix_200/tasks/data_persona_aligned_base_50_0050/verify_workplace.py CHANGED
@@ -11,19 +11,19 @@ def main() -> None:
11
  "total_score": 0,
12
  "details": [
13
  {
14
- "item": "verifier_repair_fallback",
15
  "score": 0,
16
  "max_score": 100,
17
  "passed": False,
18
- "reason": 'Original ClawBenchPro verify_workplace.py could not be recovered from a same-source valid donor. Validation reasons: syntax_error:unterminated string literal (detected at line 75):line_75.',
19
  }
20
  ],
21
- "repair_metadata": {
22
  "dataset": 'persona_aligned_mix_200',
23
  "group": 'base',
24
  "source_task_id": 'data_77',
25
  "imported_task_id": 'data_persona_aligned_base_50_0050',
26
- "repair_action": "write_conservative_zero_score_fallback",
27
  },
28
  }
29
  output_path = os.path.join(workspace, "workplace_score.json")
 
11
  "total_score": 0,
12
  "details": [
13
  {
14
+ "item": "verifier_materialization_fallback",
15
  "score": 0,
16
  "max_score": 100,
17
  "passed": False,
18
+ "reason": 'Original same-group verifier could not be materialized as an executable workplace verifier. Validation reasons: syntax_error:unterminated string literal (detected at line 75):line_75.',
19
  }
20
  ],
21
+ "verifier_materialization": {
22
  "dataset": 'persona_aligned_mix_200',
23
  "group": 'base',
24
  "source_task_id": 'data_77',
25
  "imported_task_id": 'data_persona_aligned_base_50_0050',
26
+ "action": 'conservative_fallback_raw_syntax_bad',
27
  },
28
  }
29
  output_path = os.path.join(workspace, "workplace_score.json")
persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0003/verify_workplace.py CHANGED
@@ -1,105 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
-
5
- def calculate_average_phred(qual_str):
6
- if not qual_str:
7
- return 0
8
- return sum(ord(c) - 33 for c in qual_str) / len(qual_str)
9
-
10
- def get_gold_standard(fastq_path):
11
- adapter = "GATCGGAAGAGCACACGTC"
12
- surviving_ids = []
13
-
14
- if not os.path.exists(fastq_path):
15
- return surviving_ids
16
 
17
- with open(fastq_path, 'r') as f:
18
- while True:
19
- header = f.readline().strip()
20
- if not header:
21
- break
22
- seq = f.readline().strip()
23
- plus = f.readline().strip()
24
- qual = f.readline().strip()
25
-
26
- # 规则 1: 剔除包含接头的
27
- if adapter in seq:
28
- continue
29
-
30
- # 规则 2: 平均质量分数低于 20 的剔除
31
- if calculate_average_phred(qual) < 20:
32
- continue
33
-
34
- # 提取 ID (去除开头的 @)
35
- read_id = header[1:] if header.startswith('@') else header
36
- surviving_ids.append(read_id)
37
-
38
- return surviving_ids
39
 
40
- def main():
41
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
42
- fastq_path = os.path.join(workspace, "raw_data/run_774.fastq")
43
- result_path = os.path.join(workspace, "results/surviving_reads.txt")
44
-
45
- score = 0
46
- details = []
47
-
48
- # 1. 检查结果文件是否存在
49
- if os.path.exists(result_path):
50
- score += 10
51
- details.append({"item": "结果文件存在性", "score": 10, "max_score": 10, "passed": True, "reason": "results/surviving_reads.txt 已生成"})
52
- else:
53
- details.append({"item": "结果文件存在性", "score": 0, "max_score": 10, "passed": False, "reason": "未找到 results/surviving_reads.txt"})
54
- # 如果文件不存在，后续检查无法进行
55
- with open("workplace_score.json", "w") as f:
56
- json.dump({"total_score": 0, "details": details}, f)
57
- return
58
-
59
- # 2. 读取并验证结果格式
60
- with open(result_path, 'r') as f:
61
- agent_lines = [line.strip() for line in f.readlines() if line.strip()]
62
-
63
- has_at_prefix = any(line.startswith('@') for line in agent_lines)
64
- if not has_at_prefix:
65
- score += 20
66
- details.append({"item": "输出格式正确性（无@前缀）", "score": 20, "max_score": 20, "passed": True, "reason": "Read ID 符合要求，没有包含 @ 符号"})
67
- else:
68
- details.append({"item": "输出格式正确性（无@前缀）", "score": 0, "max_score": 20, "passed": False, "reason": "部分 Read ID 仍保留了 FASTQ 的 @ 前缀"})
69
-
70
- # 3. 逻辑验证（金标准比对）
71
- gold_ids = set(get_gold_standard(fastq_path))
72
- agent_ids = set(agent_lines)
73
-
74
- # 计算交集、差集
75
- tp = len(gold_ids.intersection(agent_ids))
76
- fp = len(agent_ids - gold_ids)
77
- fn = len(gold_ids - agent_ids)
78
-
79
- if len(gold_ids) == 0:
80
- accuracy_score = 0 # 异常情况
81
- else:
82
- # 允许极小误差，但逻辑错误（如没过滤接头或质量分算错）会导致大量差异
83
- accuracy = tp / len(gold_ids) if len(gold_ids) > 0 else 0
84
- penalty = (fp / len(gold_ids)) * 0.5 # 错选惩罚
85
-
86
- final_acc_score = max(0, (accuracy - penalty) * 70)
87
- score += int(final_acc_score)
88
-
89
- if final_acc_score >= 65:
90
- details.append({"item": "数据过滤准确性", "score": int(final_acc_score), "max_score": 70, "passed": True, "reason": f"匹配率极高: 命中 {tp}, 多选 {fp}, 漏选 {fn}"})
91
- else:
92
- details.append({"item": "数据过滤准确性", "score": int(final_acc_score), "max_score": 70, "passed": False, "reason": f"过滤逻辑存在明显偏差: 命中 {tp}, 多选 {fp}, 漏选 {fn}"})
93
-
94
- # 总结输出
95
- final_score = min(100, score)
96
- output = {
97
- "total_score": int(final_score),
98
- "details": details
99
  }
100
-
101
- with open("workplace_score.json", "w") as f:
102
- json.dump(output, f, indent=2)
 
103
 
104
  if __name__ == "__main__":
105
  main()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'Original same-group verifier could not be materialized as an executable workplace verifier. Validation reasons: syntax_error:unterminated string literal (detected at line 127):line_127.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'hard',
24
+ "source_task_id": 'data_06',
25
+ "imported_task_id": 'data_persona_aligned_hard_50_0003',
26
+ "action": 'conservative_fallback_raw_syntax_bad',
27
+ },
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
  main()
persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0005/verify_workplace.py CHANGED
@@ -1,96 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
 
5
- def load_json(path):
6
- try:
7
- with open(path, 'r', encoding='utf-8') as f:
8
- return json.load(f)
9
- except Exception:
10
- return None
11
 
12
- def verify():
13
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
14
- report_path = os.path.join(workspace, "reports/termination_list.json")
15
-
16
- score = 0
17
- details = []
18
-
19
- # 1. 基础结构检查 (10分)
20
- if not os.path.exists(report_path):
21
- details.append({"item": "文件存在性", "score": 0, "max_score": 10, "passed": False, "reason": "报告文件 reports/termination_list.json 未生成"})
22
- else:
23
- data = load_json(report_path)
24
- if isinstance(data, list):
25
- score += 10
26
- details.append({"item": "JSON 格式检查", "score": 10, "max_score": 10, "passed": True, "reason": "报告为合法的 JSON 数组"})
27
- else:
28
- details.append({"item": "JSON 格式检查", "score": 0, "max_score": 10, "passed": False, "reason": "报告格式不是 JSON 数组"})
29
- data = []
30
-
31
- # 2. 核心数据解析与验证准备
32
- # 预期结果集合 (根据 env_builder 生成逻辑)
33
- # EBS: vol-0abcd111111111111 (ai-core -> alice), vol-0abcd222222222222 (data-eng -> charlie), vol-0abcd333333333333 (unknown-team -> unknown)
34
- # EC2: i-0ffff111111111111 (ai-research -> bob)
35
- expected_ebs = {"vol-0abcd111111111111", "vol-0abcd222222222222", "vol-0abcd333333333333"}
36
- expected_ec2 = {"i-0ffff111111111111"}
37
-
38
- owners_map = {
39
- "vol-0abcd111111111111": "alice.ai@mega-corp.local",
40
- "vol-0abcd222222222222": "charlie.data@mega-corp.local",
41
- "vol-0abcd333333333333": "unknown",
42
- "i-0ffff111111111111": "bob.research@mega-corp.local"
43
- }
44
-
45
- if data:
46
- # 3. EBS 资源识别 (30分)
47
- found_ebs = {item['resource_id'] for item in data if item.get('resource_type') == 'EBS'}
48
- ebs_correct = found_ebs == expected_ebs
49
- ebs_excess = found_ebs - expected_ebs # 检查是否有误杀 (如 vol-0abcd999999999999)
50
-
51
- ebs_score = 0
52
- if "vol-0abcd111111111111" in found_ebs: ebs_score += 10
53
- if "vol-0abcd222222222222" in found_ebs: ebs_score += 10
54
- if "vol-0abcd333333333333" in found_ebs: ebs_score += 10
55
- if "vol-0abcd999999999999" in found_ebs: ebs_score -= 10 # 误杀扣分
56
-
57
- ebs_score = max(0, ebs_score)
58
- score += ebs_score
59
- details.append({"item": "EBS 闲置资源识别", "score": ebs_score, "max_score": 30, "passed": ebs_score >= 30, "reason": f"识别到 EBS: {found_ebs}"})
60
-
61
- # 4. EC2 资源识别 (30分)
62
- found_ec2 = {item['resource_id'] for item in data if item.get('resource_type') == 'EC2'}
63
- ec2_score = 0
64
- if "i-0ffff111111111111" in found_ec2: ec2_score += 30
65
- if "i-0ffff222222222222" in found_ec2 or "i-0ffff333333333333" in found_ec2:
66
- ec2_score = max(0, ec2_score - 20) # 误将高负载或平均负载正常的实例列入
67
-
68
- score += ec2_score
69
- details.append({"item": "EC2 GPU低利用率识别", "score": ec2_score, "max_score": 30, "passed": ec2_score == 30, "reason": f"识别到 EC2: {found_ec2}"})
70
-
71
- # 5. Owner 匹配正确性 (30分)
72
- owner_correct_count = 0
73
- total_items = len(data)
74
- for item in data:
75
- rid = item.get('resource_id')
76
- provided_owner = item.get('owner')
77
- if rid in owners_map and provided_owner == owners_map[rid]:
78
- owner_correct_count += 1
79
-
80
- owner_score = int((owner_correct_count / total_items * 30)) if total_items > 0 else 0
81
- score += owner_score
82
- details.append({"item": "负责人邮箱交叉比对", "score": owner_score, "max_score": 30, "passed": owner_score == 30, "reason": f"正确匹配了 {owner_correct_count}/{total_items} 个负责人"})
83
-
84
- # 最终分处理
85
- final_score = min(100, max(0, score))
86
-
87
  result = {
88
- "total_score": final_score,
89
- "details": details
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
90
  }
91
-
92
- with open("workplace_score.json", "w", encoding="utf-8") as f:
93
- json.dump(result, f, indent=2, ensure_ascii=False)
 
94
 
95
  if __name__ == "__main__":
96
- verify()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
6
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
10
  result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'Original same-group verifier could not be materialized as an executable workplace verifier. Validation reasons: syntax_error:unterminated string literal (detected at line 39):line_39.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'hard',
24
+ "source_task_id": 'data_38',
25
+ "imported_task_id": 'data_persona_aligned_hard_50_0005',
26
+ "action": 'conservative_fallback_raw_syntax_bad',
27
+ },
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
+ main()
persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0008/verify_workplace.py CHANGED
@@ -1,61 +1,64 @@
1
  import os
2
  import sys
3
  import json
 
 
4
 
5
- def verify():
6
- # 基础路径处理
7
- workspace = sys.argv[1] if len(sys.argv) > 1 else "."
8
- report_path = os.path.join(workspace, "reports/bottleneck.json")
9
-
10
- score = 0
11
- details = []
12
 
13
- # 1. 检查结果文件是否存在 (10分)
14
- if os.path.exists(report_path):
15
- score += 10
16
- details.append({"item": "Check reports/bottleneck.json existence", "score": 10, "max_score": 10, "passed": True, "reason": "Report file found."})
17
- else:
18
- details.append({"item": "Check reports/bottleneck.json existence", "score": 0, "max_score": 10, "passed": False, "reason": "Report file not found."})
19
- # 如果文件不存在，后续检查无法进行，直接写入结果
20
- write_score(score, details)
21
- return
22
 
23
- # 2. 检查 JSON 格式与 Schema 合法性 (20分)
 
24
  try:
25
- with open(report_path, 'r', encoding='utf-8') as f:
26
- data = json.load(f)
27
-
28
- if "bottleneck_entity" in data:
29
- score += 20
30
- details.append({"item": "Verify JSON Schema", "score": 20, "max_score": 20, "passed": True, "reason": "Key 'bottleneck_entity' exists."})
31
-
32
- # 3. 检查 Entity ID 准确性 (70分)
33
- # 根据 env_builder.py, target_entity = "8847291"
34
- actual_id = str(data["bottleneck_entity"]).strip()
35
- expected_id = "8847291"
36
-
37
- if actual_id == expected_id:
38
- score += 70
39
- details.append({"item": "Verify Bottleneck Entity ID", "score": 70, "max_score": 70, "passed": True, "reason": f"Correct entity ID found: {actual_id}"})
40
- else:
41
- details.append({"item": "Verify Bottleneck Entity ID", "score": 0, "max_score": 70, "passed": False, "reason": f"Incorrect entity ID. Expected {expected_id}, but got {actual_id}"})
42
-
43
- else:
44
- details.append({"item": "Verify JSON Schema", "score": 0, "max_score": 20, "passed": False, "reason": "Required key 'bottleneck_entity' is missing."})
45
-
46
- except json.JSONDecodeError:
47
- details.append({"item": "Verify JSON Schema", "score": 0, "max_score": 20, "passed": False, "reason": "File is not a valid JSON."})
48
  except Exception as e:
49
- details.append({"item": "Verify JSON Schema", "score": 0, "max_score": 20, "passed": False, "reason": f"Unexpected error: {str(e)}"})
50
-
51
- write_score(score, details)
52
 
53
- def write_score(total_score, details):
54
- with open("workplace_score.json", "w", encoding="utf-8") as f:
55
- json.dump({
56
- "total_score": total_score,
57
- "details": details
58
- }, f, indent=2, ensure_ascii=False)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
59
 
60
- if __name__ == "__main__":
61
- verify()
 
1
  import os
2
  import sys
3
  import json
4
+ import httpx
5
+ from openai import OpenAI
6
 
7
+ # ---------------------------------------------------------
8
+ # [强制 API 规范]：即使当前采用严格的代码确定性验证，
9
+ # 也必须保留大模型检测基建，以备未来对非结构化文件进行校验。
10
+ # ---------------------------------------------------------
11
+ MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
12
+ MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
13
+ MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
14
 
15
+ http_client = httpx.Client(verify=False)
16
+ client = OpenAI(
17
+ api_key=MOCK_API_KEY,
18
+ base_url=MOCK_API_BASE,
19
+ http_client=http_client
20
+ )
 
 
 
21
 
22
+ def llm_judge_content(prompt_text, file_content):
23
+ """用于检测非结构化文本的统一接口"""
24
  try:
25
+ response = client.chat.completions.create(
26
+ model=MOCK_MODEL_NAME,
27
+ messages=[
28
+ {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
29
+ {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
30
+ ],
31
+ temperature=0
32
+ )
33
+ return "yes" in response.choices[0].message.content.strip().lower()
 
 
 
 
 
 
 
 
 
 
 
 
 
 
34
  except Exception as e:
35
+ print(f"LLM API Error: {e}")
36
+ return False
 
37
 
38
+ # ---------------------------------------------------------
39
+ # 核心验证逻辑：防御性编程，对格式、键值、结果准确度分级剥离
40
+ # ---------------------------------------------------------
41
+ def verify(workspace):
42
+ details = []
43
+ total_score = 0
44
+
45
+ report_path = os.path.join(workspace, "reports", "bottleneck.json")
46
+
47
+ # 考核项 1: 文件定位与留存 (10分)
48
+ if os.path.exists(report_path):
49
+ details.append({
50
+ "item": "检查目标文件是否在规范路径生成",
51
+ "score": 10, "max_score": 10, "passed": True,
52
+ "reason": "成功定位到 reports/bottleneck.json"
53
+ })
54
+ total_score += 10
55
+ else:
56
+ details.append({
57
+ "item": "检查目标文件是否在规范路径生成",
58
+ "score": 0, "max_score": 10, "passed": False,
59
+ "reason": "未能找到交差文件 reports/bottleneck.json"
60
+ })
61
+ return total_score, details
62
 
63
+ # 考核项 2: 数据结构纯净度校验 (20分)
64
+ # 防御思路：禁止返回携带 markdown 标记的冗余文本（如
persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0010/verify_workplace.py CHANGED
@@ -1,117 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import httpx
5
- from openai import OpenAI
6
-
7
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
8
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
9
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
10
 
11
- # 初始化客户端，必须关闭 SSL 验证
12
- http_client = httpx.Client(verify=False)
13
- client = OpenAI(
14
- api_key=MOCK_API_KEY,
15
- base_url=MOCK_API_BASE,
16
- http_client=http_client
17
- )
18
 
19
- def llm_judge_content(prompt_text, file_content):
20
- # 此函数为检测非结构化文本的统一接口
21
- try:
22
- response = client.chat.completions.create(
23
- model=MOCK_MODEL_NAME,
24
- messages=[
25
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
26
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
27
- ],
28
- temperature=0
29
- )
30
- return "yes" in response.choices[0].message.content.strip().lower()
31
- except Exception as e:
32
- print(f"LLM API Error: {e}")
33
- return False
34
-
35
- def verify():
36
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
37
- target_file = os.path.join(workspace, "report", "failed_init.json")
38
-
39
- total_score = 0
40
- details = []
41
-
42
- # 1. 检查目标文件是否存在 (20分)
43
- if os.path.isfile(target_file):
44
- total_score += 20
45
- details.append({"item": "检查结果文件是否存在", "score": 20, "max_score": 20, "passed": True, "reason": "文件 report/failed_init.json 存在"})
46
- else:
47
- details.append({"item": "检查结果文件是否存在", "score": 0, "max_score": 20, "passed": False, "reason": "文件 report/failed_init.json 不存在"})
48
- write_score(total_score, details)
49
- return
50
-
51
- # 2. 检查文件是否为合法的 JSON 格式 (20分)
52
- try:
53
- with open(target_file, 'r', encoding='utf-8') as f:
54
- data = json.load(f)
55
- total_score += 20
56
- details.append({"item": "检查文件是否为合法 JSON", "score": 20, "max_score": 20, "passed": True, "reason": "成功解析 JSON 文件"})
57
- except json.JSONDecodeError:
58
- details.append({"item": "检查文件是否为合法 JSON", "score": 0, "max_score": 20, "passed": False, "reason": "文件内容不是合法的 JSON 格式"})
59
- write_score(total_score, details)
60
- return
61
- except Exception as e:
62
- details.append({"item": "检查文件是否为合法 JSON", "score": 0, "max_score": 20, "passed": False, "reason": f"文件读取发生未知错误: {str(e)}"})
63
- write_score(total_score, details)
64
- return
65
-
66
- # 3. 检查 JSON 字段完整性 (10分)
67
- if not isinstance(data, dict):
68
- details.append({"item": "检查 JSON 结构类型", "score": 0, "max_score": 10, "passed": False, "reason": "JSON 根节点必须是一个对象 (dict)"})
69
- write_score(total_score, details)
70
- return
71
-
72
- has_register = "register" in data
73
- has_value = "value" in data
74
- extra_keys = set(data.keys()) - {"register", "value"}
75
-
76
- if has_register and has_value:
77
- if extra_keys:
78
- # 存在冗余字段，扣5分
79
- total_score += 5
80
- details.append({"item": "检查 JSON 字段", "score": 5, "max_score": 10, "passed": False, "reason": f"包含了必要的字段，但存在冗余字段: {extra_keys}"})
81
- else:
82
- total_score += 10
83
- details.append({"item": "检查 JSON 字段", "score": 10, "max_score": 10, "passed": True, "reason": "精确包含了 register 和 value 字段"})
84
- else:
85
- details.append({"item": "检查 JSON 字段", "score": 0, "max_score": 10, "passed": False, "reason": "缺失 register 或 value 字段"})
86
- write_score(total_score, details)
87
- return
88
-
89
- # 4. 提取并校验 register 的值 (25分)
90
- reg_val = str(data.get("register", "")).strip().lower()
91
- if reg_val == "0x2a":
92
- total_score += 25
93
- details.append({"item": "校验报错的寄存器地址", "score": 25, "max_score": 25, "passed": True, "reason": "找到了正确的寄存器地址 0x2A"})
94
- else:
95
- details.append({"item": "校验报错的寄存器地址", "score": 0, "max_score": 25, "passed": False, "reason": f"寄存器地址错误或格式不标准，预期 '0x2A'，实际获取到 '{reg_val}'"})
96
-
97
- # 5. 提取并校验 value 的值 (25分)
98
- val_val = str(data.get("value", "")).strip().lower()
99
- if val_val == "0x7f":
100
- total_score += 25
101
- details.append({"item": "校验试图写入的错误数据", "score": 25, "max_score": 25, "passed": True, "reason": "找到了正确的报错数据 0x7F"})
102
- else:
103
- details.append({"item": "校验试图写入的错误数据", "score": 0, "max_score": 25, "passed": False, "reason": f"试图写入的数据错误或格式不标准，预期 '0x7F'，实际获取到 '{val_val}'"})
104
-
105
- write_score(total_score, details)
106
-
107
- def write_score(total_score, details):
108
- report = {
109
- "total_score": total_score,
110
- "details": details
111
  }
112
- with open("workplace_score.json", "w", encoding="utf-8") as f:
113
- json.dump(report, f, indent=2, ensure_ascii=False)
114
- print(json.dumps(report, indent=2, ensure_ascii=False))
 
115
 
116
  if __name__ == "__main__":
117
- verify()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'Original same-group verifier could not be materialized as an executable workplace verifier. Validation reasons: syntax_error:unterminated string literal (detected at line 55):line_55.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'hard',
24
+ "source_task_id": 'data_59',
25
+ "imported_task_id": 'data_persona_aligned_hard_50_0010',
26
+ "action": 'conservative_fallback_raw_syntax_bad',
27
+ },
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
+ main()
persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0018/verify_workplace.py CHANGED
@@ -1,142 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import re
5
- import httpx
6
- from openai import OpenAI
7
-
8
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
9
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
10
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
11
 
12
- # 初始化客户端，必须关闭 SSL 验证
13
- http_client = httpx.Client(verify=False)
14
- client = OpenAI(
15
- api_key=MOCK_API_KEY,
16
- base_url=MOCK_API_BASE,
17
- http_client=http_client
18
- )
19
 
20
- def llm_judge_content(prompt_text, file_content):
21
- # 此函数为检测非结构化文本的统一接口
22
- try:
23
- response = client.chat.completions.create(
24
- model=MOCK_MODEL_NAME,
25
- messages=[
26
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
27
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
28
- ],
29
- temperature=0
30
- )
31
- return "yes" in response.choices[0].message.content.strip().lower()
32
- except Exception as e:
33
- print(f"LLM API Error: {e}")
34
- return False
35
-
36
- def get_ground_truth(workspace):
37
- """
38
- 沙盒内探针自己执行硬核逻辑，重新解析数据计算 Ground Truth，避免假阴性/假阳性。
39
- """
40
- can_log = os.path.join(workspace, "chassis_can.log")
41
- radar_json = os.path.join(workspace, "sensor_data", "radar_track.json")
42
-
43
- if not os.path.exists(can_log) or not os.path.exists(radar_json):
44
- return set()
45
-
46
- aeb_timestamps = []
47
- with open(can_log, "r", encoding="utf-8") as f:
48
- for line in f:
49
- # 必须满足双重条件: 刹车CAN ID = 0x2B0, PAYLOAD 前两字节 = FF 01
50
- if "MSG_ID:0x2B0" in line and "PAYLOAD:[FF 01" in line:
51
- m = re.search(r"<(\d+)>", line)
52
- if m:
53
- aeb_timestamps.append(int(m.group(1)))
54
-
55
- truth_ids = set()
56
- with open(radar_json, "r", encoding="utf-8") as f:
57
- radar_data = json.load(f)
58
-
59
- frames = radar_data.get("data_stream", {}).get("radar_front_center", {}).get("frames", [])
60
- for frame in frames:
61
- stamp_ms = frame.get("header", {}).get("stamp_ms", 0)
62
- # 严密的时间戳对齐：雷达比底盘快 1500ms
63
- if (stamp_ms - 1500) in aeb_timestamps:
64
- objects = frame.get("payload", {}).get("tracked_entities", {}).get("radar_objects", [])
65
- for obj in objects:
66
- rcs = obj.get("attributes", {}).get("rcs_dbsm", 999.0)
67
- conf = obj.get("attributes", {}).get("track_confidence", 999)
68
- # 必须满足 rcs < 5.0 且 confidence < 60
69
- if rcs < 5.0 and conf < 60:
70
- tid = obj.get("metadata", {}).get("track_id", "")
71
- if tid:
72
- truth_ids.add(tid)
73
-
74
- return truth_ids
75
-
76
- def main():
77
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
78
- target_file = os.path.join(workspace, "analysis", "ghost_ids.json")
79
-
80
- details = []
81
- total_score = 0
82
-
83
- # 1. 验证目标文件存在性 (10分)
84
- if os.path.exists(target_file):
85
- total_score += 10
86
- details.append({"item": "检查目标文件是否存在", "score": 10, "max_score": 10, "passed": True, "reason": "文件 analysis/ghost_ids.json 存在"})
87
- else:
88
- details.append({"item": "检查目标文件是否存在", "score": 0, "max_score": 10, "passed": False, "reason": "文件 analysis/ghost_ids.json 不存在"})
89
-
90
- # 2. 验证结构纯净性 (20分)
91
- # 绝对禁止使用正则去匹配结构化结果，必须使用 json 库严格解析
92
- agent_ids = []
93
- is_valid_format = False
94
- if os.path.exists(target_file):
95
- try:
96
- with open(target_file, "r", encoding="utf-8") as f:
97
- data = json.load(f)
98
- if isinstance(data, list) and all(isinstance(i, str) for i in data):
99
- is_valid_format = True
100
- agent_ids = data
101
- total_score += 20
102
- details.append({"item": "JSON格式规范性验证", "score": 20, "max_score": 20, "passed": True, "reason": "是一个纯净的字符串数组"})
103
- else:
104
- details.append({"item": "JSON格式规范性验证", "score": 0, "max_score": 20, "passed": False, "reason": "结构错误，不是纯净的字符串数组"})
105
- except json.JSONDecodeError:
106
- details.append({"item": "JSON格式规范性验证", "score": 0, "max_score": 20, "passed": False, "reason": "非法的JSON文件"})
107
- else:
108
- details.append({"item": "JSON格式规范性验证", "score": 0, "max_score": 20, "passed": False, "reason": "文件缺失，无法验证"})
109
-
110
- # 3. 数据精准度 (70分)
111
- if is_valid_format:
112
- truth_ids = get_ground_truth(workspace)
113
- agent_set = set(agent_ids)
114
-
115
- if not truth_ids:
116
- # 如果极端情况环境加载异常，这里进行容错
117
- details.append({"item": "验证提取的 ID 准确性", "score": 0, "max_score": 70, "passed": False, "reason": "Ground Truth 数据生成错误，请检查环境"})
118
- else:
119
- intersection = agent_set.intersection(truth_ids)
120
- false_positives = agent_set - truth_ids
121
- false_negatives = truth_ids - agent_set
122
-
123
- union_len = len(agent_set.union(truth_ids))
124
- # 使用严格的 Jaccard 相似度来反映 F1 维度的惩罚机制，有捏造、漏报均会急剧降分
125
- data_score = int(70 * (len(intersection) / union_len)) if union_len > 0 else 0
126
-
127
- total_score += data_score
128
- passed = (data_score == 70)
129
- reason = f"精准度检查完成。正确提取: {len(intersection)}项, 漏报: {len(false_negatives)}项, 误报(幻觉/条件错误): {len(false_positives)}项"
130
- details.append({"item": "验证提取的 ID 准确性", "score": data_score, "max_score": 70, "passed": passed, "reason": reason})
131
- else:
132
- details.append({"item": "验证提取的 ID 准确性", "score": 0, "max_score": 70, "passed": False, "reason": "由于文件不存在或格式不合规，跳过数据校验"})
133
 
134
- # 统分写入
135
- with open(os.path.join(workspace, "workplace_score.json"), "w", encoding="utf-8") as f:
136
- json.dump({
137
- "total_score": total_score,
138
- "details": details
139
- }, f, indent=2, ensure_ascii=False)
140
 
141
  if __name__ == "__main__":
142
  main()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'Original same-group verifier could not be materialized as an executable workplace verifier. Validation reasons: missing_score_output_marker; syntax_error:unterminated string literal (detected at line 119):line_119.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'hard',
24
+ "source_task_id": 'data_08',
25
+ "imported_task_id": 'data_persona_aligned_hard_50_0018',
26
+ "action": 'conservative_fallback_raw_syntax_bad',
27
+ },
28
+ }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
32
 
 
 
 
 
 
 
33
 
34
  if __name__ == "__main__":
35
  main()
persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0025/verify_workplace.py CHANGED
@@ -1,129 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import httpx
5
- from openai import OpenAI
6
 
7
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
8
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
9
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
10
 
11
- # 初始化客户端，必须关闭 SSL 验证
12
- http_client = httpx.Client(verify=False)
13
- client = OpenAI(
14
- api_key=MOCK_API_KEY,
15
- base_url=MOCK_API_BASE,
16
- http_client=http_client
17
- )
18
-
19
- def llm_judge_content(prompt_text, file_content):
20
- # 此函数为检测非结构化文本的统一接口
21
- try:
22
- response = client.chat.completions.create(
23
- model=MOCK_MODEL_NAME,
24
- messages=[
25
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
26
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
27
- ],
28
- temperature=0
29
- )
30
- return "yes" in response.choices[0].message.content.strip().lower()
31
- except Exception as e:
32
- print(f"LLM API Error: {e}")
33
- return False
34
-
35
- def main():
36
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
37
- target_file = os.path.join(workspace, "risk_control", "blacklist.json")
38
-
39
- score = 0
40
- details = []
41
-
42
- # 1. 检查目标目录及文件是否存在 (10 分)
43
- if os.path.exists(target_file):
44
- score += 10
45
- details.append({"item": "检查目标文件是否存在", "score": 10, "max_score": 10, "passed": True, "reason": "risk_control/blacklist.json 存在"})
46
- else:
47
- details.append({"item": "检查目标文件是否存在", "score": 0, "max_score": 10, "passed": False, "reason": "未找到 risk_control/blacklist.json"})
48
- with open("workplace_score.json", "w", encoding="utf-8") as f:
49
- json.dump({"total_score": score, "details": details}, f, ensure_ascii=False, indent=2)
50
- return
51
-
52
- # 2. 检查 JSON 格式合法性 (15 分)
53
- try:
54
- with open(target_file, "r", encoding="utf-8") as f:
55
- data = json.load(f)
56
- score += 15
57
- details.append({"item": "检查 JSON 格式合法性", "score": 15, "max_score": 15, "passed": True, "reason": "成功解析为合法 JSON 格式"})
58
- except Exception as e:
59
- details.append({"item": "检查 JSON 格式合法性", "score": 0, "max_score": 15, "passed": False, "reason": f"JSON 解析失败: {e}"})
60
- with open("workplace_score.json", "w", encoding="utf-8") as f:
61
- json.dump({"total_score": score, "details": details}, f, ensure_ascii=False, indent=2)
62
- return
63
-
64
- # 确保根节点是字典
65
- if not isinstance(data, dict):
66
- details.append({"item": "检查 JSON 根节点类型", "score": 0, "max_score": 75, "passed": False, "reason": "JSON 根节点必须是对象(字典)"})
67
- with open("workplace_score.json", "w", encoding="utf-8") as f:
68
- json.dump({"total_score": score, "details": details}, f, ensure_ascii=False, indent=2)
69
- return
70
-
71
- # 定位并验证键名 (大小写不敏感，但必须是正确的 FIX 字段)
72
- clordid_key = None
73
- sender_key = None
74
- for k in data.keys():
75
- kl = k.lower()
76
- if kl == "clordid":
77
- clordid_key = k
78
- elif kl == "sendercompid":
79
- sender_key = k
80
-
81
- # 3. 验证 ClOrdID 键 (10 分)
82
- if clordid_key:
83
- score += 10
84
- details.append({"item": "验证 ClOrdID 键是否存在", "score": 10, "max_score": 10, "passed": True, "reason": f"找到规范键名: {clordid_key}"})
85
- else:
86
- details.append({"item": "验证 ClOrdID 键是否存在", "score": 0, "max_score": 10, "passed": False, "reason": "未找到符合 ClOrdID 的键名"})
87
-
88
- # 4. 验证 SenderCompID 键 (10 分)
89
- if sender_key:
90
- score += 10
91
- details.append({"item": "验证 SenderCompID 键是否存在", "score": 10, "max_score": 10, "passed": True, "reason": f"找到规范键名: {sender_key}"})
92
- else:
93
- details.append({"item": "验证 SenderCompID 键是否存在", "score": 0, "max_score": 10, "passed": False, "reason": "未找到符合 SenderCompID 的键名"})
94
-
95
- # 5. 结构与幻觉检查 (10 分)
96
- if len(data.keys()) == 2 and clordid_key and sender_key:
97
- score += 10
98
- details.append({"item": "验证是否无多余字段 (防幻觉)", "score": 10, "max_score": 10, "passed": True, "reason": "字段数量严格为 2，未捏造多余信息"})
99
- else:
100
- details.append({"item": "验证是否无多余字段 (防幻觉)", "score": 0, "max_score": 10, "passed": False, "reason": f"存在多余字段或结构错误，当前包含的键: {list(data.keys())}"})
101
-
102
- # 6. 验证 ClOrdID 数值准确性 (20 分)
103
- if clordid_key:
104
- val = str(data[clordid_key]).strip()
105
- if val == "POISON_HFT_0x9A":
106
- score += 20
107
- details.append({"item": "验��� ClOrdID 提取准确性", "score": 20, "max_score": 20, "passed": True, "reason": "ClOrdID 提取准确无误"})
108
- else:
109
- details.append({"item": "验证 ClOrdID 提取准确性", "score": 0, "max_score": 20, "passed": False, "reason": f"目标数据提取错误，当前值: {val}"})
110
- else:
111
- details.append({"item": "验证 ClOrdID 提取准确性", "score": 0, "max_score": 20, "passed": False, "reason": "因缺少对应键名而跳过判定"})
112
-
113
- # 7. 验证 SenderCompID 数值准确性 (25 分)
114
- if sender_key:
115
- val = str(data[sender_key]).strip()
116
- if val == "MALICIOUS_VIP_88":
117
- score += 25
118
- details.append({"item": "验证 SenderCompID 提取准确性", "score": 25, "max_score": 25, "passed": True, "reason": "SenderCompID 提取准确无误"})
119
- else:
120
- details.append({"item": "验证 SenderCompID 提取准确性", "score": 0, "max_score": 25, "passed": False, "reason": f"目标数据提取错误，当前值: {val}"})
121
- else:
122
- details.append({"item": "验证 SenderCompID 提取准确性", "score": 0, "max_score": 25, "passed": False, "reason": "因缺少对应键名而跳过判定"})
123
 
124
- # 输出最终成绩报告
125
- with open("workplace_score.json", "w", encoding="utf-8") as f:
126
- json.dump({"total_score": score, "details": details}, f, ensure_ascii=False, indent=2)
127
 
128
  if __name__ == "__main__":
129
  main()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
6
 
 
 
 
7
 
8
+ def main() -> None:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'Original same-group verifier could not be materialized as an executable workplace verifier. Validation reasons: syntax_error:unterminated string literal (detected at line 162):line_162.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'hard',
24
+ "source_task_id": 'data_25',
25
+ "imported_task_id": 'data_persona_aligned_hard_50_0025',
26
+ "action": 'conservative_fallback_raw_syntax_bad',
27
+ },
28
+ }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
32
 
 
 
 
33
 
34
  if __name__ == "__main__":
35
  main()
persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0026/verify_workplace.py CHANGED
@@ -1,170 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import httpx
5
- from openai import OpenAI
6
-
7
- # ==========================================
8
- # 强制 API 规范：大模型初始化
9
- # ==========================================
10
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
11
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
12
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
13
 
14
- # 初始化客户端，必须关闭 SSL 验证
15
- http_client = httpx.Client(verify=False)
16
- client = OpenAI(
17
- api_key=MOCK_API_KEY,
18
- base_url=MOCK_API_BASE,
19
- http_client=http_client
20
- )
21
 
22
- def llm_judge_content(prompt_text, file_content):
23
- """
24
- 此函数为检测非结构化文本的统一接口。
25
- 针对 data_persona_aligned_hard_50_0026 任务由于要求生成纯 JSON 结果，代码可以直接完成确定性验证，
26
- 本函数作为规范预留，在有任何自然语言生成时调用。
27
- """
28
- try:
29
- response = client.chat.completions.create(
30
- model=MOCK_MODEL_NAME,
31
- messages=[
32
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
33
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
34
- ],
35
- temperature=0
36
- )
37
- return "yes" in response.choices[0].message.content.strip().lower()
38
- except Exception as e:
39
- print(f"LLM API Error: {e}")
40
- return False
41
-
42
- def verify():
43
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
44
- results = []
45
- total_score = 0
46
-
47
- # 1. 动态计算 Ground Truth (由于 env_builder 中数据和 hex 值是随机生成的)
48
- ground_truth = None
49
- traces_dir = os.path.join(workspace, "traces")
50
- if os.path.isdir(traces_dir):
51
- for file_name in os.listdir(traces_dir):
52
- if not file_name.endswith(".json"):
53
- continue
54
- file_path = os.path.join(traces_dir, file_name)
55
- try:
56
- with open(file_path, 'r', encoding='utf-8') as f:
57
- data = json.load(f)
58
- for trace in data.get("data", []):
59
- spans = trace.get("spans", [])
60
- is_target = False
61
-
62
- # 找到持续时间 > 5s (5,000,000 微秒) 的目标 Trace
63
- for span in spans:
64
- if span.get("duration", 0) > 5000000:
65
- is_target = True
66
- break
67
-
68
- if is_target:
69
- trace_id = trace.get("traceID")
70
- operation = None
71
- payload = None
72
- # 深度遍历锁定底层报错 Span
73
- for span in spans:
74
- if span.get("operationName") == "grpc.inventory.ReserveStock":
75
- operation = span.get("operationName")
76
- for log in span.get("logs", []):
77
- for field in log.get("fields", []):
78
- if field.get("key") == "corrupted_payload":
79
- payload = field.get("value")
80
-
81
- if trace_id and operation and payload:
82
- ground_truth = {
83
- "trace_id": trace_id,
84
- "operation": operation,
85
- "payload": payload
86
- }
87
- break
88
- except Exception:
89
- pass
90
-
91
- if ground_truth:
92
- break
93
-
94
- if not ground_truth:
95
- results.append({"item": "沙盒异常监控", "score": 0, "max_score": 0, "passed": False, "reason": "无法计算 Ground Truth，Trace 原始文件可能遭到破坏或未生成。"})
96
- with open(os.path.join(workspace, "workplace_score.json"), "w", encoding='utf-8') as f:
97
- json.dump({"total_score": 0, "details": results}, f, indent=2, ensure_ascii=False)
98
- return
99
-
100
- # 2. 检查结果文件是否存在 (权重: 20分)
101
- target_file = os.path.join(workspace, "ops", "root_cause.json")
102
- if not os.path.exists(target_file):
103
- results.append({"item": "检查目标文件是否存在", "score": 0, "max_score": 20, "passed": False, "reason": "文件 ops/root_cause.json 不存在"})
104
- with open(os.path.join(workspace, "workplace_score.json"), "w", encoding='utf-8') as f:
105
- json.dump({"total_score": 0, "details": results}, f, indent=2, ensure_ascii=False)
106
- return
107
-
108
- results.append({"item": "检查目标文件是否存在", "score": 20, "max_score": 20, "passed": True, "reason": "文件 ops/root_cause.json 存在"})
109
- total_score += 20
110
-
111
- # 3. 检查文件是否为合法 JSON 且结构正常 (权重: 20分)
112
- try:
113
- with open(target_file, 'r', encoding='utf-8') as f:
114
- ans_data = json.load(f)
115
- results.append({"item": "检查文件是否为合法 JSON 解析", "score": 20, "max_score": 20, "passed": True, "reason": "标准 JSON 格式合法"})
116
- total_score += 20
117
- except json.JSONDecodeError:
118
- results.append({"item": "检查文件是否为合法 JSON 解析", "score": 0, "max_score": 20, "passed": False, "reason": "无法被原生 json.load 解析"})
119
- with open(os.path.join(workspace, "workplace_score.json"), "w", encoding='utf-8') as f:
120
- json.dump({"total_score": total_score, "details": results}, f, indent=2, ensure_ascii=False)
121
- return
122
- except Exception as e:
123
- results.append({"item": "检查文件是否为合法 JSON 解析", "score": 0, "max_score": 20, "passed": False, "reason": f"文件读取发生未知错误: {str(e)}"})
124
- with open(os.path.join(workspace, "workplace_score.json"), "w", encoding='utf-8') as f:
125
- json.dump({"total_score": total_score, "details": results}, f, indent=2, ensure_ascii=False)
126
- return
127
-
128
- # 4. 提取核心指标：严格比对 (每项 20 分，共 60 分)
129
-
130
- # 4.1 Trace ID 校验
131
- agent_trace_id = ans_data.get("trace_id")
132
- if agent_trace_id == ground_truth["trace_id"]:
133
- results.append({"item": "检查 Trace ID 提取是否正确", "score": 20, "max_score": 20, "passed": True, "reason": "Trace ID 精准匹配"})
134
- total_score += 20
135
- else:
136
- results.append({"item": "检查 Trace ID 提取是否正确", "score": 0, "max_score": 20, "passed": False, "reason": f"匹配失败, 期待 {ground_truth['trace_id']}, 实际得到 {agent_trace_id}"})
137
-
138
- # 4.2 底层 Operation 校验
139
- agent_operation = ans_data.get("operation")
140
- if agent_operation == ground_truth["operation"]:
141
- results.append({"item": "检查底层报错 Operation 提取是否正确", "score": 20, "max_score": 20, "passed": True, "reason": "Operation 提取正确"})
142
- total_score += 20
143
- else:
144
- results.append({"item": "检查底层报错 Operation 提取是否正确", "score": 0, "max_score": 20, "passed": False, "reason": f"匹配失败, 实际得到 {agent_operation}"})
145
-
146
- # 4.3 Corrupted Payload 内存地址校验
147
- agent_payload = ans_data.get("payload")
148
- if agent_payload == ground_truth["payload"]:
149
- results.append({"item": "检查 Corrupted Payload 提取是否正确", "score": 20, "max_score": 20, "passed": True, "reason": "Payload 提取正确"})
150
- total_score += 20
151
- else:
152
- results.append({"item": "检查 Corrupted Payload 提取是否正确", "score": 0, "max_score": 20, "passed": False, "reason": f"匹配失败, 实际得到 {agent_payload}"})
153
-
154
- # 5. 扣分项：极其严格地验证废话与多余字段 (题目要求："其他废话和分析过程一句都别留")
155
- allowed_keys = {"trace_id", "operation", "payload"}
156
- actual_keys = set(ans_data.keys())
157
- extra_keys = actual_keys - allowed_keys
158
- if extra_keys:
159
- deduct = 20
160
- total_score = max(0, total_score - deduct)
161
- results.append({"item": "多余废话字段检测", "score": -deduct, "max_score": 0, "passed": False, "reason": f"存在不允许的额外字段: {extra_keys}，违背强制不罗嗦指令，扣除 {deduct} 分"})
162
- else:
163
- results.append({"item": "多余废话字段检测", "score": 0, "max_score": 0, "passed": True, "reason": "未包含多余字段，严格遵守了输出格式指令"})
164
 
165
- # 最终输出 workplace_score.json
166
- with open(os.path.join(workspace, "workplace_score.json"), "w", encoding='utf-8') as f:
167
- json.dump({"total_score": total_score, "details": results}, f, indent=2, ensure_ascii=False)
168
 
169
  if __name__ == "__main__":
170
- verify()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'Original same-group verifier could not be materialized as an executable workplace verifier. Validation reasons: syntax_error:unterminated string literal (detected at line 143):line_143.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'hard',
24
+ "source_task_id": 'data_48',
25
+ "imported_task_id": 'data_persona_aligned_hard_50_0026',
26
+ "action": 'conservative_fallback_raw_syntax_bad',
27
+ },
28
+ }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
32
 
 
 
 
33
 
34
  if __name__ == "__main__":
35
+ main()
persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0029/verify_workplace.py CHANGED
@@ -1,172 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import httpx
5
- from openai import OpenAI
6
-
7
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
8
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
9
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
10
 
11
- # 初始化客户端，必须关闭 SSL 验证
12
- http_client = httpx.Client(verify=False)
13
- client = OpenAI(
14
- api_key=MOCK_API_KEY,
15
- base_url=MOCK_API_BASE,
16
- http_client=http_client
17
- )
18
 
19
- def llm_judge_content(prompt_text, file_content):
20
- # 此函数为检测非结构化文本的统一接口
21
- try:
22
- response = client.chat.completions.create(
23
- model=MOCK_MODEL_NAME,
24
- messages=[
25
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
26
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
27
- ],
28
- temperature=0
29
- )
30
- return "yes" in response.choices[0].message.content.strip().lower()
31
- except Exception as e:
32
- print(f"LLM API Error: {e}")
33
- return False
34
-
35
- def main():
36
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
37
- total_score = 0
38
- details = []
39
-
40
- target_file = os.path.join(workspace, "action_items", "kill_list.json")
41
-
42
- # 1. 检查目标文件是否存在 (10 分)
43
- if os.path.exists(target_file):
44
- details.append({
45
- "item": "检查结果文件是否存在",
46
- "score": 10,
47
- "max_score": 10,
48
- "passed": True,
49
- "reason": "目标文件 action_items/kill_list.json 已创建"
50
- })
51
- total_score += 10
52
- else:
53
- details.append({
54
- "item": "检查结果文件是否存在",
55
- "score": 0,
56
- "max_score": 10,
57
- "passed": False,
58
- "reason": "目标文件 action_items/kill_list.json 未找到"
59
- })
60
- # 文件不存在直接输出结果
61
- with open("workplace_score.json", "w", encoding="utf-8") as f:
62
- json.dump({"total_score": total_score, "details": details}, f, ensure_ascii=False, indent=2)
63
- return
64
-
65
- # 2. 检查 JSON 格式合法性与 Schema (20 分)
66
- # 利用原生的 json.load 严查 Markdown 包裹、废话及格式错误
67
- data = None
68
- try:
69
- with open(target_file, "r", encoding="utf-8") as f:
70
- data = json.load(f)
71
-
72
- if isinstance(data, dict) and "idle_ebs" in data and "zombie_gpu" in data:
73
- if isinstance(data["idle_ebs"], list) and isinstance(data["zombie_gpu"], list):
74
- details.append({
75
- "item": "检查 JSON 格式与 Schema 合法性",
76
- "score": 20,
77
- "max_score": 20,
78
- "passed": True,
79
- "reason": "JSON 文件可以被原生解析器成功加载，没有包含多余的废话和 Markdown 代码块，且 Schema 正确"
80
- })
81
- total_score += 20
82
- else:
83
- details.append({
84
- "item": "检查 JSON 格式与 Schema 合法性",
85
- "score": 0,
86
- "max_score": 20,
87
- "passed": False,
88
- "reason": "JSON 格式有效，但 idle_ebs 或 zombie_gpu 不是列表"
89
- })
90
- data = None
91
- else:
92
- details.append({
93
- "item": "检查 JSON 格式与 Schema 合法性",
94
  "score": 0,
95
- "max_score": 20,
96
  "passed": False,
97
- "reason": "JSON 格式有效，但缺少要求的 idle_ebs 或 zombie_gpu 字段"
98
- })
99
- data = None
100
- except json.JSONDecodeError as e:
101
- details.append({
102
- "item": "检查 JSON 格式与 Schema 合法性",
103
- "score": 0,
104
- "max_score": 20,
105
- "passed": False,
106
- "reason": f"JSON 解析失败（Agent 未遵循要求，可能包裹了 Markdown、包含了废话说明或语法错误）：{str(e)}"
107
- })
108
-
109
- # 如果无法解析，后续计分均跳过
110
- if data:
111
- # 定义期望的答案集
112
- expected_ebs = {"vol-09a8b7c6d5e4f3a21", "vol-00001111222233334", "vol-0ffeeddccbbaa9988"}
113
- # 定义一定存在于文件中但不应该被提取的干扰项（用于校验是否存在提取条件过滤错误）
114
- invalid_ebs = {"vol-01122334455667788", "vol-0a1b2c3d4e5f60708"}
115
-
116
- expected_gpu = {"i-0987654321abcdef0", "i-55556666777788889", "i-deadbeefdeadbeef0", "i-9876543210fedcba9"}
117
- invalid_gpu = {"i-11112222333344445", "i-99990000aaaaabbbb", "i-abcdef12345678900"}
118
-
119
- actual_ebs_set = set(data.get("idle_ebs", []))
120
- actual_gpu_set = set(data.get("zombie_gpu", []))
121
-
122
- # 3. 检查 idle_ebs 提取准确度 (满分 35 分)
123
- ebs_score = 0
124
- ebs_reason = ""
125
-
126
- # 严查作弊与逻辑错误：一旦包含了不符合条件的数据或幻觉伪造数据，一票否决
127
- if any(x in invalid_ebs for x in actual_ebs_set) or not actual_ebs_set.issubset(expected_ebs | invalid_ebs):
128
- ebs_reason = "在 idle_ebs 结果中混入了 in-use 的 EBS 或无中生有的幻觉 ID，触发强杀脚本报警规则，该项得分清零。"
129
- else:
130
- if "vol-09a8b7c6d5e4f3a21" in actual_ebs_set: ebs_score += 10
131
- if "vol-00001111222233334" in actual_ebs_set: ebs_score += 10
132
- if "vol-0ffeeddccbbaa9988" in actual_ebs_set: ebs_score += 15 # 提取单引号伪 JSON 数据的难度稍高
133
- ebs_reason = f"成功提取了 {len(actual_ebs_set)} 个符合要求的可用 EBS 卷。"
134
-
135
- details.append({
136
- "item": "检查 idle_ebs 数据准确性",
137
- "score": ebs_score,
138
- "max_score": 35,
139
- "passed": ebs_score == 35,
140
- "reason": ebs_reason
141
- })
142
- total_score += ebs_score
143
-
144
- # 4. 检查 zombie_gpu 提取准确度 (满分 35 分)
145
- gpu_score = 0
146
- gpu_reason = ""
147
-
148
- # 同样严查：如果提取出利用率大于2%的节点，或者把非GPU实例拿进来，一票否决
149
- if any(x in invalid_gpu for x in actual_gpu_set) or not actual_gpu_set.issubset(expected_gpu | invalid_gpu):
150
- gpu_reason = "在 zombie_gpu 中包含了利用率大于2%的实例、非 GPU 实例(如t3)或幻觉 ID，触发报警，该项得分清零。"
151
- else:
152
- if "i-0987654321abcdef0" in actual_gpu_set: gpu_score += 8
153
- if "i-55556666777788889" in actual_gpu_set: gpu_score += 9
154
- if "i-deadbeefdeadbeef0" in actual_gpu_set: gpu_score += 9
155
- if "i-9876543210fedcba9" in actual_gpu_set: gpu_score += 9
156
- gpu_reason = f"成功提取了 {len(actual_gpu_set)} 个符合要求的僵尸 GPU 实例。"
157
-
158
- details.append({
159
- "item": "检查 zombie_gpu 数据准确性",
160
- "score": gpu_score,
161
- "max_score": 35,
162
- "passed": gpu_score == 35,
163
- "reason": gpu_reason
164
- })
165
- total_score += gpu_score
166
 
167
- # 输出结果
168
- with open("workplace_score.json", "w", encoding="utf-8") as f:
169
- json.dump({"total_score": total_score, "details": details}, f, ensure_ascii=False, indent=2)
170
 
171
  if __name__ == "__main__":
172
  main()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
15
  "score": 0,
16
+ "max_score": 100,
17
  "passed": False,
18
+ "reason": 'Original same-group verifier could not be materialized as an executable workplace verifier. Validation reasons: syntax_error:unterminated string literal (detected at line 53):line_53.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'hard',
24
+ "source_task_id": 'data_18',
25
+ "imported_task_id": 'data_persona_aligned_hard_50_0029',
26
+ "action": 'conservative_fallback_raw_syntax_bad',
27
+ },
28
+ }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
32
 
 
 
 
33
 
34
  if __name__ == "__main__":
35
  main()
persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0031/verify_workplace.py CHANGED
@@ -1,130 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import httpx
5
- from openai import OpenAI
6
-
7
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
8
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
9
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
10
-
11
- # 初始化客户端，强制关闭 SSL 验证
12
- http_client = httpx.Client(verify=False)
13
- client = OpenAI(
14
- api_key=MOCK_API_KEY,
15
- base_url=MOCK_API_BASE,
16
- http_client=http_client
17
- )
18
 
19
- def llm_judge_content(prompt_text, file_content):
20
- """用于处理非结构化破坏情况的备用大模型兜底探针"""
21
- try:
22
- response = client.chat.completions.create(
23
- model=MOCK_MODEL_NAME,
24
- messages=[
25
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
26
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
27
- ],
28
- temperature=0
29
- )
30
- return "yes" in response.choices[0].message.content.strip().lower()
31
- except Exception as e:
32
- print(f"LLM API Error: {e}")
33
- return False
34
 
35
- def verify_workplace():
36
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
37
- score_details = []
38
- total_score = 0
39
-
40
- blacklist_path = os.path.join(workspace, "config", "blacklist.json")
41
-
42
- # Target IPS derived from deterministic environment builder
43
- target_ips = {"120.44.55.66", "45.33.22.11", "10.0.5.200"}
44
-
45
- if os.path.exists(blacklist_path):
46
- score_details.append({"item": "检查目标文件是否存在", "score": 10, "max_score": 10, "passed": True, "reason": "文件 config/blacklist.json 已生成"})
47
- total_score += 10
48
-
49
- try:
50
- with open(blacklist_path, 'r', encoding='utf-8') as f:
51
- content = f.read()
52
-
53
- # 严格代码解析结构化数据
54
- data = json.loads(content)
55
- score_details.append({"item": "检查文件是否符合标准 JSON Schema", "score": 10, "max_score": 10, "passed": True, "reason": "解析器成功加载 JSON"})
56
- total_score += 10
57
-
58
- if isinstance(data, list) and all(isinstance(x, str) for x in data):
59
- score_details.append({"item": "检查数据结构是否为纯粹的字符串数组", "score": 10, "max_score": 10, "passed": True, "reason": "文件顶层确认为包含字符串的 List，未夹带私货字段"})
60
- total_score += 10
61
-
62
- # 去重检查
63
- extracted_ips = set(data)
64
- if len(data) == len(extracted_ips) and len(data) > 0:
65
- score_details.append({"item": "检查数据是否去重", "score": 10, "max_score": 10, "passed": True, "reason": "源 IP 无冗余和重复"})
66
- total_score += 10
67
- else:
68
- score_details.append({"item": "检查数据是否去重", "score": 0, "max_score": 10, "passed": False, "reason": "数组包含重复元素或为空"})
69
-
70
- # 严密准确度计算（满分 60分）
71
- correct_count = len(extracted_ips.intersection(target_ips))
72
- wrong_count = len(extracted_ips - target_ips)
73
-
74
- accuracy_score = correct_count * 20 # 找到一个给 20 分
75
- accuracy_score -= wrong_count * 15 # 抓错一个幻觉扣 15 分
76
-
77
- if extracted_ips == target_ips:
78
- accuracy_score = 60
79
- else:
80
- accuracy_score = max(0, min(accuracy_score, 59))
81
-
82
- if accuracy_score == 60:
83
- score_details.append({"item": "源 IP 提取准确度评估", "score": 60, "max_score": 60, "passed": True, "reason": "极其精准！捕获了全部 3 个 MALFORMED 源 IP，且没有任何干扰项或幻觉"})
84
- elif accuracy_score > 0:
85
- score_details.append({"item": "源 IP 提取准确度评估", "score": accuracy_score, "max_score": 60, "passed": False, "reason": f"提取了部分准确数据（{correct_count} 命中），但遗漏或掺杂了非攻击 IP 幻觉（{wrong_count} 错误）"})
86
- else:
87
- score_details.append({"item": "源 IP 提取准确度评估", "score": 0, "max_score": 60, "passed": False, "reason": "未能提取有效恶意 IP，或幻觉伪造内容过多导致准确度清零"})
88
-
89
- total_score += accuracy_score
90
-
91
- else:
92
- score_details.append({"item": "检查数据结构是否为纯粹的字符串数组", "score": 0, "max_score": 10, "passed": False, "reason": "JSON 内部结构并非单纯的字符串数组"})
93
-
94
- # LLM 非结构化数据挽回机制
95
- prompt = "Does the following text contain ALL three of these specific IP addresses: '120.44.55.66', '45.33.22.11', '10.0.5.200'?"
96
- if llm_judge_content(prompt, content):
97
- score_details.append({"item": "LLM 语义兜底提取 (容错机制)", "score": 30, "max_score": 70, "passed": False, "reason": "Agent 输出了非合规的 JSON 结构被代码解析器拦截，但 LLM 判定其内部包含了全部目标恶意 IP，发放部分容错辛苦分"})
98
- total_score += 30
99
- else:
100
- score_details.append({"item": "LLM 语义兜底提取 (容错机制)", "score": 0, "max_score": 70, "passed": False, "reason": "格式错误，且未能正确分析出全部的恶意源 IP"})
101
-
102
- except json.JSONDecodeError:
103
- score_details.append({"item": "检查文件是否符合标准 JSON Schema", "score": 0, "max_score": 10, "passed": False, "reason": "原生解析失败，文件可能包含解释性前缀、Markdown 块或其他非 JSON 格式废话"})
104
- score_details.append({"item": "检查数据结构是否为纯粹的字符串数组", "score": 0, "max_score": 10, "passed": False, "reason": "前置条件失败"})
105
-
106
- # LLM 非结构化数据挽回机制
107
- with open(blacklist_path, 'r', encoding='utf-8') as f:
108
- content = f.read()
109
- prompt = "Does the following text contain ALL three of these specific IP addresses: '120.44.55.66', '45.33.22.11', '10.0.5.200'?"
110
- if llm_judge_content(prompt, content):
111
- score_details.append({"item": "LLM 语义兜底提取 (容错机制)", "score": 30, "max_score": 70, "passed": False, "reason": "文件彻底损坏，但大模型判定其中散落了所有的目标源 IP，发放部分容错辛苦分"})
112
- total_score += 30
113
- else:
114
- score_details.append({"item": "LLM 语义兜底提取 (容错机制)", "score": 0, "max_score": 70, "passed": False, "reason": "文件彻底损坏，且错失了目标恶意源 IP"})
115
-
116
- else:
117
- score_details.append({"item": "检查目标文件是否存在", "score": 0, "max_score": 10, "passed": False, "reason": f"关键交付物 {blacklist_path} 缺失"})
118
- score_details.append({"item": "结构与精准度验证", "score": 0, "max_score": 90, "passed": False, "reason": "基础文件丢失，取消剩余考核链条"})
119
-
120
- # 结果强制写回
121
  result = {
122
- "total_score": total_score,
123
- "details": score_details
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
124
  }
125
-
126
- with open(os.path.join(workspace, "workplace_score.json"), "w", encoding="utf-8") as f:
127
- json.dump(result, f, ensure_ascii=False, indent=2)
 
128
 
129
  if __name__ == "__main__":
130
- verify_workplace()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
10
  result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'Original same-group verifier could not be materialized as an executable workplace verifier. Validation reasons: syntax_error:unterminated string literal (detected at line 72):line_72.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'hard',
24
+ "source_task_id": 'data_66',
25
+ "imported_task_id": 'data_persona_aligned_hard_50_0031',
26
+ "action": 'conservative_fallback_raw_syntax_bad',
27
+ },
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
+ main()
persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0039/verify_workplace.py CHANGED
@@ -1,151 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import httpx
5
- import subprocess
6
- from openai import OpenAI
7
-
8
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
9
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
10
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
11
 
12
- # 初始化客户端，必须关闭 SSL 验证
13
- http_client = httpx.Client(verify=False)
14
- client = OpenAI(
15
- api_key=MOCK_API_KEY,
16
- base_url=MOCK_API_BASE,
17
- http_client=http_client
18
- )
19
 
20
- def llm_judge_content(prompt_text, file_content):
21
- try:
22
- response = client.chat.completions.create(
23
- model=MOCK_MODEL_NAME,
24
- messages=[
25
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
26
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
27
- ],
28
- temperature=0
29
- )
30
- return "yes" in response.choices[0].message.content.strip().lower()
31
- except Exception as e:
32
- print(f"LLM API Error: {e}")
33
- return False
34
-
35
- def get_ground_truth_files(workspace_dir):
36
- # 动态尝试通过特需 Skill 获取 Ground Truth (防止硬编码死锁)
37
- gt_files = []
38
- # 首先尝试相对于 workspace 的路径
39
- skill_path = os.path.join(workspace_dir, "skills", "data_persona_aligned_hard_50_0039", "ext4_inode_query_skill.py")
40
- if not os.path.exists(skill_path):
41
- # Fallback：尝试相对于评测脚本当前执行环境的路径
42
- skill_path = os.path.abspath(os.path.join(os.path.dirname(__file__), "../../skills/data_persona_aligned_hard_50_0039/ext4_inode_query_skill.py"))
43
-
44
- if os.path.exists(skill_path):
45
- try:
46
- for inode in [1024, 50000, 99999, 12, 8888]:
47
- res = subprocess.run([sys.executable, skill_path, str(inode)], capture_output=True, text=True, timeout=2)
48
- if res.returncode == 0:
49
- gt_files.append(res.stdout.strip())
50
- except Exception:
51
- pass
52
- return gt_files
53
-
54
- def main():
55
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
56
- plan_path = os.path.join(workspace, "recovery_plan.json")
57
-
58
- total_score = 0
59
- details = []
60
-
61
- # 1. 结构与存在性检测 (15分)
62
- if not os.path.exists(plan_path):
63
- details.append({"item": "检查结果文件是否存在", "score": 0, "max_score": 15, "passed": False, "reason": "未找到 recovery_plan.json"})
64
- write_score(workspace, 0, details)
65
- return
66
-
67
- try:
68
- with open(plan_path, "r", encoding="utf-8") as f:
69
- plan_data = json.load(f)
70
- except Exception as e:
71
- details.append({"item": "检查JSON格式与结构合法性", "score": 0, "max_score": 15, "passed": False, "reason": f"JSON解析失败: {e}"})
72
- write_score(workspace, 0, details)
73
- return
74
-
75
- expected_keys = {"crash_source_line", "lost_files"}
76
- actual_keys = set(plan_data.keys())
77
- if actual_keys != expected_keys:
78
- details.append({
79
- "item": "检查JSON格式与结构合法性", "score": 0, "max_score": 15, "passed": False,
80
- "reason": f"包含多余或缺少字段，预期 {expected_keys}，实际 {actual_keys}。严惩捏造幻觉！"
81
- })
82
- else:
83
- details.append({"item": "检查JSON格式与结构合法性", "score": 15, "max_score": 15, "passed": True, "reason": "字段完全一致"})
84
- total_score += 15
85
-
86
- # 2. 纯代码严谨结构校验：数组数量与类型 (25分)
87
- lost_files = plan_data.get("lost_files", [])
88
- if not isinstance(lost_files, list):
89
- details.append({"item": "校验 lost_files 数据类型", "score": 0, "max_score": 25, "passed": False, "reason": "lost_files 不是数组结构"})
90
- elif len(lost_files) != 5:
91
- details.append({"item": "校验提取的文件数量精确度", "score": 0, "max_score": 25, "passed": False, "reason": f"应当精确提取5个文件，实际提取了 {len(lost_files)} 个"})
92
- else:
93
- is_all_strs = all(isinstance(x, str) for x in lost_files)
94
- has_no_raw_digits = all(not str(x).isdigit() for x in lost_files)
95
- if is_all_strs and has_no_raw_digits:
96
- details.append({"item": "校验提取的文件数量与基础类型", "score": 25, "max_score": 25, "passed": True, "reason": "成功提取出5个合法字符串节点，未直接填入原始 Inode 数字"})
97
- total_score += 25
98
- else:
99
- details.append({"item": "校验提取的文件数量与基础类型", "score": 5, "max_score": 25, "passed": False, "reason": "包含非字符串或纯数字（可能直接填入了 inode 未调用恢复工具）"})
100
- total_score += 5
101
-
102
- # 3. 业务文件溯源准确度 - 结合 GT 精确比对 (30分)
103
- gt_files = get_ground_truth_files(workspace)
104
- if len(gt_files) == 5:
105
- # 有确定的 Ground Truth，执行极其严苛的精准比对
106
- matched = len(set(lost_files).intersection(set(gt_files)))
107
- if matched == 5:
108
- details.append({"item": "核对恢复业务文件名精确度", "score": 30, "max_score": 30, "passed": True, "reason": "5个业务文件名与工具底层 Ground Truth 完全一致"})
109
- total_score += 30
110
- else:
111
- details.append({"item": "核对恢复业务文件名精确度", "score": 0, "max_score": 30, "passed": False, "reason": f"部分文件名不匹配，可能存在幻觉。仅正确 {matched}/5"})
112
- else:
113
- # 降级验证 (如果在特定容器中由于权限无法拉起 skill)
114
- if isinstance(lost_files, list) and len(lost_files) == 5 and all('/' in str(x) or '.' in str(x) for x in lost_files):
115
- details.append({"item": "核对恢复业务文件名精确度(降级模式)", "score": 30, "max_score": 30, "passed": True, "reason": "文件名格式符合绝对路径特征"})
116
- total_score += 30
117
- else:
118
- details.append({"item": "核对恢复业务文件名精确度(降级模式)", "score": 0, "max_score": 30, "passed": False, "reason": "未通过路径结构特征校验"})
119
-
120
- # 4. 利用 LLM 进行非结构化/半结构化自然语义判定 - 捕捉被陷阱误导的回答 (30分)
121
- crash_line = plan_data.get("crash_source_line", "")
122
- if not crash_line or not isinstance(crash_line, str):
123
- details.append({"item": "大模型判定源代码行解析结果", "score": 0, "max_score": 30, "passed": False, "reason": "未提供或类型非字符串"})
124
- else:
125
- prompt = (
126
- "The agent was asked to resolve a Linux kernel RIP address ('ffffffff812ab340') into a human-readable crash source code line "
127
- "using a specific fallback lookup skill. The primary tool is a TRAP and returns a 'missing symbol error'. "
128
- "Evaluate if the string provided strictly contains a successful C source file path and line number "
129
- "(e.g., 'fs/ext4/orphan.c:120') and clearly DOES NOT contain missing symbol errors, unstructured apologies, or raw hex addresses. "
130
- "It must reflect a successful technical resolution."
131
- )
132
- passed = llm_judge_content(prompt, crash_line)
133
- if passed:
134
- details.append({"item": "大模型判定源代码行解析结果", "score": 30, "max_score": 30, "passed": True, "reason": "大模型判定内容精确反映了已翻译的源代码行，避开了陷阱工具的干扰"})
135
- total_score += 30
136
- else:
137
- details.append({"item": "大模型判定源代码行解析结果", "score": 0, "max_score": 30, "passed": False, "reason": "大模型判定解析失败：可能是原始16进制、踩中了陷阱工具报错，或存在严重幻觉冗余"})
138
-
139
- write_score(workspace, total_score, details)
140
-
141
- def write_score(workspace, total_score, details):
142
- output_data = {
143
- "total_score": total_score,
144
- "details": details
145
  }
146
- score_path = os.path.join(workspace, "workplace_score.json")
147
- with open(score_path, "w", encoding="utf-8") as f:
148
- json.dump(output_data, f, indent=2, ensure_ascii=False)
 
149
 
150
  if __name__ == "__main__":
151
  main()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'Original same-group verifier could not be materialized as an executable workplace verifier. Validation reasons: syntax_error:unterminated string literal (detected at line 66):line_66.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'hard',
24
+ "source_task_id": 'data_93',
25
+ "imported_task_id": 'data_persona_aligned_hard_50_0039',
26
+ "action": 'conservative_fallback_raw_syntax_bad',
27
+ },
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
  main()
persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0041/verify_workplace.py CHANGED
@@ -1,100 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import re
5
-
6
- def verify():
7
- workspace = sys.argv[1] if len(sys.argv) > 1 else "."
8
- score = 0
9
- details = []
10
-
11
- ioc_file_path = os.path.join(workspace, "report/ioc.json")
12
-
13
- # 1. 检查文件是否存在 (10分)
14
- if os.path.exists(ioc_file_path):
15
- score += 10
16
- details.append({"item": "检查 report/ioc.json 物理文件", "score": 10, "max_score": 10, "passed": True, "reason": "文件已生成"})
17
-
18
- # 2. 检查 JSON 格式合法性 (10分)
19
- try:
20
- with open(ioc_file_path, 'r', encoding='utf-8') as f:
21
- data = json.load(f)
22
- score += 10
23
- details.append({"item": "检查 JSON 语法合法性", "score": 10, "max_score": 10, "passed": True, "reason": "JSON 解析成功"})
24
-
25
- # 3. 检查持久化恶意路径 (40分)
26
- # 期待路径: C:\ProgramData\Intel\telemetry_srv.exe
27
- expected_path = r"C:\ProgramData\Intel\telemetry_srv.exe".lower()
28
-
29
- found_path = False
30
- actual_path_val = ""
31
- # 遍历 JSON 寻找类似路径的字符串
32
- def find_path_recursive(obj):
33
- nonlocal found_path, actual_path_val
34
- if isinstance(obj, str):
35
- if "telemetry_srv.exe" in obj.lower():
36
- actual_path_val = obj
37
- if obj.lower().replace("/", "\\") == expected_path:
38
- found_path = True
39
- elif isinstance(obj, dict):
40
- for v in obj.values(): find_path_recursive(v)
41
- elif isinstance(obj, list):
42
- for i in obj: find_path_recursive(i)
43
 
44
- find_path_recursive(data)
45
-
46
- if found_path:
47
- score += 40
48
- details.append({"item": "恶意程序持久化路径匹配", "score": 40, "max_score": 40, "passed": True, "reason": f"成功提取并匹配路径: {actual_path_val}"})
49
- elif "telemetry_srv.exe" in actual_path_val.lower():
50
- score += 20
51
- details.append({"item": "恶意程序持久化路径匹配", "score": 20, "max_score": 40, "passed": False, "reason": f"提取到的路径不完整或不准确: {actual_path_val}"})
52
- else:
53
- details.append({"item": "恶意程序持久化路径匹配", "score": 0, "max_score": 40, "passed": False, "reason": "未找到正确的恶意程序路径"})
54
 
55
- # 4. 检查脱壳特征码 (40分)
56
- # 期待特征码: 5C7A8E1F2B3D4C5A6B7C8D9EAFB0C1D2 (16字节)
57
- expected_sig = "5c7a8e1f2b3d4c5a6b7c8d9eafb0c1d2"
58
-
59
- found_sig = False
60
- actual_sig_val = ""
61
-
62
- def find_sig_recursive(obj):
63
- nonlocal found_sig, actual_sig_val
64
- if isinstance(obj, str):
65
- # 移除空格、破折号、0x等干扰项进行匹配
66
- clean_str = re.sub(r'[^a-fA-F0-9]', '', obj).lower()
67
- if expected_sig in clean_str:
68
- actual_sig_val = obj
69
- found_sig = True
70
- elif isinstance(obj, dict):
71
- for v in obj.values(): find_sig_recursive(v)
72
- elif isinstance(obj, list):
73
- for i in obj: find_sig_recursive(i)
74
-
75
- find_sig_recursive(data)
76
-
77
- if found_sig:
78
- score += 40
79
- details.append({"item": "16字节脱壳特征码匹配", "score": 40, "max_score": 40, "passed": True, "reason": f"成功匹配特征码: {actual_sig_val}"})
80
- else:
81
- details.append({"item": "16字节脱壳特征码匹配", "score": 0, "max_score": 40, "passed": False, "reason": "未找到或特征码错误"})
82
-
83
- except json.JSONDecodeError:
84
- details.append({"item": "检查 JSON 语法合法性", "score": 0, "max_score": 10, "passed": False, "reason": "JSON 格式错误，无法解析"})
85
- details.append({"item": "数据内容检查", "score": 0, "max_score": 80, "passed": False, "reason": "由于 JSON 无法解析，跳过内容检查"})
86
- else:
87
- details.append({"item": "检查 report/ioc.json 物理文件", "score": 0, "max_score": 10, "passed": False, "reason": "文件未生成"})
88
- details.append({"item": "后续所有检查", "score": 0, "max_score": 90, "passed": False, "reason": "基础文件缺失"})
89
-
90
- # 最终输出结果
91
  result = {
92
- "total_score": score,
93
- "details": details
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
94
  }
95
-
96
- with open("workplace_score.json", "w", encoding="utf-8") as f:
97
- json.dump(result, f, indent=2, ensure_ascii=False)
 
98
 
99
  if __name__ == "__main__":
100
- verify()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
9
+ workspace = sys.argv[1] if len(sys.argv) > 1 else "."
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
10
  result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'Original same-group verifier could not be materialized as an executable workplace verifier. Validation reasons: empty; missing_score_output_marker.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'hard',
24
+ "source_task_id": 'data_84',
25
+ "imported_task_id": 'data_persona_aligned_hard_50_0041',
26
+ "action": 'conservative_fallback_raw_empty',
27
+ },
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
+ main()
persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0042/verify_workplace.py CHANGED
@@ -1,135 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import re
5
- import httpx
6
- from openai import OpenAI
7
-
8
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
9
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
10
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
11
 
12
- # 初始化客户端，必须关闭 SSL 验证
13
- http_client = httpx.Client(verify=False)
14
- client = OpenAI(
15
- api_key=MOCK_API_KEY,
16
- base_url=MOCK_API_BASE,
17
- http_client=http_client
18
- )
19
 
20
- def llm_judge_content(prompt_text, file_content):
21
- # 此函数为检测非结构化文本的统一接口
22
- try:
23
- response = client.chat.completions.create(
24
- model=MOCK_MODEL_NAME,
25
- messages=[
26
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
27
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
28
- ],
29
- temperature=0
30
- )
31
- return "yes" in response.choices[0].message.content.strip().lower()
32
- except Exception as e:
33
- print(f"LLM API Error: {e}")
34
- return False
35
-
36
- def normalize_hex_string(s):
37
- """去除干扰字符，纯化十六进制用于宽松比对"""
38
- return re.sub(r'[^A-Fa-f0-9]', '', str(s)).upper()
39
-
40
- def verify():
41
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
42
- json_path = os.path.join(workspace, "analysis", "dirty_tx.json")
43
-
44
- details = []
45
- total_score = 0
46
-
47
- # 1. 检查目标目录和文件是否存在 (10 分)
48
- if os.path.exists(json_path):
49
- details.append({"item": "检查目标文件 dirty_tx.json 是否存在", "score": 10, "max_score": 10, "passed": True, "reason": "文件存在"})
50
- total_score += 10
51
- else:
52
- details.append({"item": "检查目标文件 dirty_tx.json 是否存在", "score": 0, "max_score": 10, "passed": False, "reason": "文件不存在"})
53
- return write_result(total_score, details)
54
-
55
- # 2. 解析 JSON 文件格式 (10 分)
56
- try:
57
- with open(json_path, "r", encoding="utf-8") as f:
58
- data = json.load(f)
59
- if isinstance(data, dict):
60
- details.append({"item": "检查 JSON 格式是否为字典", "score": 10, "max_score": 10, "passed": True, "reason": "JSON 解析成功且根结构为字典"})
61
- total_score += 10
62
- else:
63
- details.append({"item": "检查 JSON 格式是否为字典", "score": 0, "max_score": 10, "passed": False, "reason": f"根结构不是字典，类型为 {type(data)}"})
64
- return write_result(total_score, details)
65
- except Exception as e:
66
- details.append({"item": "检查 JSON 格式是否为字典", "score": 0, "max_score": 10, "passed": False, "reason": f"JSON 解析失败: {e}"})
67
- return write_result(total_score, details)
68
-
69
- # 3. 检查 Transaction ID 过滤逻辑 (30 分)
70
- expected_keys = {"TX-1002", "TX-1008"}
71
- wrong_key_0c4 = "TX-1003"
72
- actual_keys = set(data.keys())
73
-
74
- if actual_keys == expected_keys:
75
- details.append({"item": "检查提取的 Transaction ID 集合", "score": 30, "max_score": 30, "passed": True, "reason": "精确提取了触发 0C7 的异常 ID，没有多余或遗漏"})
76
- total_score += 30
77
- else:
78
- if wrong_key_0c4 in actual_keys:
79
- details.append({"item": "检查提取的 Transaction ID 集合", "score": 0, "max_score": 30, "passed": False, "reason": "混入了触发 0C4 的 TX-1003 或其他非法字段，严重违反业务规则一票否决该项"})
80
- elif expected_keys.issubset(actual_keys):
81
- details.append({"item": "检查提取的 Transaction ID 集合", "score": 0, "max_score": 30, "passed": False, "reason": "包含了不应存在的捏造键，判定为幻觉或提取逻辑错误"})
82
- else:
83
- correct_cnt = len(actual_keys.intersection(expected_keys))
84
- score_for_keys = correct_cnt * 10
85
- details.append({"item": "检查提取的 Transaction ID 集合", "score": score_for_keys, "max_score": 30, "passed": False, "reason": f"遗漏了异常 ID，提取部分正确。当前键: {actual_keys}"})
86
- total_score += score_for_keys
87
-
88
- # 4. 检查 TX-1002 的 Hex 数据段提取 (25 分)
89
- if "TX-1002" in data:
90
- expected_hex_1002 = "E3E760F1F0F0F20000012A4C40404040"
91
- actual_raw = str(data["TX-1002"])
92
- actual_hex = normalize_hex_string(actual_raw)
93
- if actual_hex == expected_hex_1002:
94
- # 进一步检查是否"保留空格"
95
- if len(actual_raw.split()) == 16:
96
- details.append({"item": "校验 TX-1002 数据段准确性", "score": 25, "max_score": 25, "passed": True, "reason": "十六进制值准确无误，且保留了空格分割"})
97
- total_score += 25
98
- else:
99
- details.append({"item": "校验 TX-1002 数据段准确性", "score": 20, "max_score": 25, "passed": True, "reason": "十六进制数值准确，但未严格按要求保留源数据的空格结构"})
100
- total_score += 20
101
- else:
102
- details.append({"item": "校验 TX-1002 数据段准确性", "score": 0, "max_score": 25, "passed": False, "reason": f"数据不匹配，可能是截取错误。找到的内容: {actual_hex}"})
103
- else:
104
- details.append({"item": "校验 TX-1002 数据段准确性", "score": 0, "max_score": 25, "passed": False, "reason": "字典中缺失 TX-1002 键"})
105
-
106
- # 5. 检查 TX-1008 的 Hex 数据段提取 (25 分)
107
- if "TX-1008" in data:
108
- expected_hex_1008 = "E3E760F1F0F0F80000FFFFFC40404040"
109
- actual_raw = str(data["TX-1008"])
110
- actual_hex = normalize_hex_string(actual_raw)
111
- if actual_hex == expected_hex_1008:
112
- if len(actual_raw.split()) == 16:
113
- details.append({"item": "校验 TX-1008 数据段准确性", "score": 25, "max_score": 25, "passed": True, "reason": "十六进制值准确无误，且保留了空格分割"})
114
- total_score += 25
115
- else:
116
- details.append({"item": "校验 TX-1008 数据段准确性", "score": 20, "max_score": 25, "passed": True, "reason": "十六进制数值准确，但未严格按要求保留源数据的空格结构"})
117
- total_score += 20
118
- else:
119
- details.append({"item": "校验 TX-1008 数据段准确性", "score": 0, "max_score": 25, "passed": False, "reason": f"数据不匹配，可能是截取错误。找到的内容: {actual_hex}"})
120
- else:
121
- details.append({"item": "校验 TX-1008 数据段准确性", "score": 0, "max_score": 25, "passed": False, "reason": "字典中缺失 TX-1008 键"})
122
-
123
- return write_result(total_score, details)
124
-
125
- def write_result(total_score, details):
126
  result = {
127
- "total_score": total_score,
128
- "details": details
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
129
  }
130
- with open("workplace_score.json", "w", encoding="utf-8") as f:
131
- json.dump(result, f, indent=2, ensure_ascii=False)
132
- return result
 
133
 
134
  if __name__ == "__main__":
135
- verify()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
10
  result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'Original same-group verifier could not be materialized as an executable workplace verifier. Validation reasons: empty; missing_score_output_marker.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'hard',
24
+ "source_task_id": 'data_74',
25
+ "imported_task_id": 'data_persona_aligned_hard_50_0042',
26
+ "action": 'conservative_fallback_raw_empty',
27
+ },
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
+ main()
persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0044/verify_workplace.py CHANGED
@@ -11,19 +11,19 @@ def main() -> None:
11
  "total_score": 0,
12
  "details": [
13
  {
14
- "item": "verifier_repair_fallback",
15
  "score": 0,
16
  "max_score": 100,
17
  "passed": False,
18
- "reason": 'Original ClawBenchPro verify_workplace.py could not be recovered from a same-source valid donor. Validation reasons: syntax_error:unterminated string literal (detected at line 92):line_92.',
19
  }
20
  ],
21
- "repair_metadata": {
22
  "dataset": 'persona_aligned_mix_200',
23
  "group": 'hard',
24
  "source_task_id": 'data_63',
25
  "imported_task_id": 'data_persona_aligned_hard_50_0044',
26
- "repair_action": "write_conservative_zero_score_fallback",
27
  },
28
  }
29
  output_path = os.path.join(workspace, "workplace_score.json")
 
11
  "total_score": 0,
12
  "details": [
13
  {
14
+ "item": "verifier_materialization_fallback",
15
  "score": 0,
16
  "max_score": 100,
17
  "passed": False,
18
+ "reason": 'Original same-group verifier could not be materialized as an executable workplace verifier. Validation reasons: syntax_error:unterminated string literal (detected at line 92):line_92.',
19
  }
20
  ],
21
+ "verifier_materialization": {
22
  "dataset": 'persona_aligned_mix_200',
23
  "group": 'hard',
24
  "source_task_id": 'data_63',
25
  "imported_task_id": 'data_persona_aligned_hard_50_0044',
26
+ "action": 'conservative_fallback_raw_syntax_bad',
27
  },
28
  }
29
  output_path = os.path.join(workspace, "workplace_score.json")
persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0045/verify_workplace.py CHANGED
@@ -1,180 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import httpx
5
- from openai import OpenAI
6
 
7
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
8
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
9
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
10
 
11
- # 初始化客户端，必须关闭 SSL 验证
12
- http_client = httpx.Client(verify=False)
13
- client = OpenAI(
14
- api_key=MOCK_API_KEY,
15
- base_url=MOCK_API_BASE,
16
- http_client=http_client
17
- )
18
-
19
- def llm_judge_content(prompt_text, file_content):
20
- # 此函数为检测非结构化文本的统一接口
21
- try:
22
- response = client.chat.completions.create(
23
- model=MOCK_MODEL_NAME,
24
- messages=[
25
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
26
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
27
- ],
28
- temperature=0
29
- )
30
- return "yes" in response.choices[0].message.content.strip().lower()
31
- except Exception as e:
32
- print(f"LLM API Error: {e}")
33
- return False
34
-
35
- def verify():
36
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
37
- score_details = []
38
- total_score = 0
39
-
40
- report_path = os.path.join(workspace, "incident_report", "culprit.json")
41
-
42
- # -------------------------------------------------------------
43
- # 检查点 1: 结果文件是否存在 (10分)
44
- # -------------------------------------------------------------
45
- item1 = {"item": "检查目标结果文件 culprit.json 是否存在", "max_score": 10, "score": 0, "passed": False, "reason": ""}
46
- if os.path.exists(report_path):
47
- item1["score"] = 10
48
- item1["passed"] = True
49
- item1["reason"] = "找到了 culprit.json 文件"
50
- else:
51
- item1["reason"] = "未找到 culprit.json 文件"
52
- score_details.append(item1)
53
-
54
- if not item1["passed"]:
55
- # 如果文件不存在，后续检查无法进行，直接输出 0 分
56
- with open("workplace_score.json", "w", encoding="utf-8") as f:
57
- json.dump({"total_score": 0, "details": score_details}, f, indent=2, ensure_ascii=False)
58
- return
59
-
60
- # -------------------------------------------------------------
61
- # 检查点 2: 严格合法的 JSON 格式及字段结构 (20分)
62
- # -------------------------------------------------------------
63
- item2 = {"item": "检查 JSON 格式及键值是否严格符合要求（严查幻觉捏造字段）", "max_score": 20, "score": 0, "passed": False, "reason": ""}
64
- data = {}
65
- try:
66
- with open(report_path, "r", encoding="utf-8") as f:
67
- data = json.load(f)
68
- keys = set(data.keys())
69
- expected_keys = {"namespace", "pod_name", "owner_team"}
70
-
71
- if keys == expected_keys:
72
- item2["score"] = 20
73
- item2["passed"] = True
74
- item2["reason"] = "JSON解析成功，且严格包含所需三个键段，无冗余"
75
- elif expected_keys.issubset(keys):
76
- item2["score"] = 10
77
- item2["reason"] = "JSON解析成功，包含所需键，但捏造了多余的键（存在轻微幻觉或未遵守严格规范）"
78
- else:
79
- missing = expected_keys - keys
80
- item2["reason"] = f"JSON结构不完整，缺少必要的键: {missing}"
81
- except json.JSONDecodeError:
82
- item2["reason"] = "文件内容不是合法的JSON格式"
83
- except Exception as e:
84
- item2["reason"] = f"读取或解析文件时发生错误: {str(e)}"
85
- score_details.append(item2)
86
-
87
- # 提取提取字段准备数值验证
88
- namespace = str(data.get("namespace", "")).strip()
89
- pod_name = str(data.get("pod_name", "")).strip()
90
- owner_team = str(data.get("owner_team", "")).strip()
91
-
92
- # -------------------------------------------------------------
93
- # 检查点 3: 精准匹配 namespace (15分)
94
- # -------------------------------------------------------------
95
- item3 = {"item": "验证 namespace 精准提取结果", "max_score": 15, "score": 0, "passed": False, "reason": ""}
96
- if namespace == "finance-production":
97
- item3["score"] = 15
98
- item3["passed"] = True
99
- item3["reason"] = "正确识别并提取了 finance-production"
100
- else:
101
- item3["reason"] = f"namespace 错误: 期望 finance-production, 实际为 '{namespace}'"
102
- score_details.append(item3)
103
-
104
- # -------------------------------------------------------------
105
- # 检查点 4: 精准匹配 pod_name (25分 - 核心难点)
106
- # -------------------------------------------------------------
107
- item4 = {"item": "验证 pod_name 精准提取结果", "max_score": 25, "score": 0, "passed": False, "reason": ""}
108
- if pod_name == "core-payment-gateway-deployment-78dbb9c4":
109
- item4["score"] = 25
110
- item4["passed"] = True
111
- item4["reason"] = "准确无误地找出了出事 Pod 名���"
112
- elif pod_name == "core-payment-gateway-deployment":
113
- item4["score"] = 10
114
- item4["reason"] = "提取了 Deployment 名称而非具体的 Pod 名称，扣除部分分数"
115
- else:
116
- item4["reason"] = f"pod_name 错误: 实际得到 '{pod_name}'"
117
- score_details.append(item4)
118
-
119
- # -------------------------------------------------------------
120
- # 检查点 5: 精准匹配 owner_team (15分)
121
- # -------------------------------------------------------------
122
- item5 = {"item": "验证 owner_team 精准提取结果", "max_score": 15, "score": 0, "passed": False, "reason": ""}
123
- if owner_team == "billing-core-team":
124
- item5["score"] = 15
125
- item5["passed"] = True
126
- item5["reason"] = "正确从杂乱的 YAML 中提取了 owner_team 标注"
127
- else:
128
- item5["reason"] = f"owner_team 错误: 实际得到 '{owner_team}'"
129
- score_details.append(item5)
130
-
131
- # -------------------------------------------------------------
132
- # 检查点 6: LLM 语义校验 —— 严禁写长篇大论 (15分)
133
- # -------------------------------------------------------------
134
- item6 = {"item": "大模型语义校验：检查是否服从了“别给我写长篇大论”的Persona指令", "max_score": 15, "score": 15, "passed": True, "reason": "未发现冗余的长篇大论文件和自然语言解释"}
135
-
136
- # 1. 扫描目录下是否存在多余的分析文件
137
- incident_dir = os.path.join(workspace, "incident_report")
138
- extra_files = []
139
- if os.path.exists(incident_dir):
140
- for root, dirs, files in os.walk(incident_dir):
141
- for file in files:
142
- if file != "culprit.json":
143
- extra_files.append(os.path.join(root, file))
144
-
145
- violation = False
146
- if extra_files:
147
- for ef in extra_files:
148
- try:
149
- with open(ef, "r", encoding="utf-8", errors="ignore") as f:
150
- content = f.read(3000)
151
- # 提示词：如果内容是在写报告、写原因分析、对话，则视为冗余的长篇大论
152
- prompt = "Check if this content is a conversational analysis report, chatty explanation, or verbose debugging details. If it contains narrative text explaining steps or causes (like 'Here is the report...', 'The reason is...'), answer YES. If it's pure raw data/empty, answer NO."
153
- if llm_judge_content(prompt, content):
154
- violation = True
155
- item6["reason"] = f"大模型判定文件 {os.path.basename(ef)} 是长篇大论的分析报告，严重违反了用户的禁止指令"
156
- break
157
- except Exception:
158
- pass
159
-
160
- if violation:
161
- item6["score"] = 0
162
- item6["passed"] = False
163
-
164
- # 2. 检查 JSON 文件内部的值是否被 Agent 加入了对话式的幻觉文本（比如 "The pod name is xxx"）
165
- if item6["passed"]:
166
- prompt2 = "Check if the following JSON structure contains conversational or narrative text in its values (e.g. 'I found the pod...', 'Team is billing...'). If it contains conversational language, answer YES. If it strictly contains pure entity names/IDs only, answer NO."
167
- if llm_judge_content(prompt2, json.dumps(data, ensure_ascii=False)):
168
- item6["score"] = 0
169
- item6["passed"] = False
170
- item6["reason"] = "大模型判定 JSON 的值中夹带了口语化解释或对话文本，未能严格提供纯净的机器数据"
171
-
172
- score_details.append(item6)
173
 
174
- # 计算总分并输出
175
- total_score = sum(d["score"] for d in score_details)
176
- with open("workplace_score.json", "w", encoding="utf-8") as f:
177
- json.dump({"total_score": total_score, "details": score_details}, f, indent=2, ensure_ascii=False)
178
 
179
  if __name__ == "__main__":
180
- verify()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
6
 
 
 
 
7
 
8
+ def main() -> None:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'Original same-group verifier could not be materialized as an executable workplace verifier. Validation reasons: syntax_error:unterminated string literal (detected at line 74):line_74.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'hard',
24
+ "source_task_id": 'data_76',
25
+ "imported_task_id": 'data_persona_aligned_hard_50_0045',
26
+ "action": 'conservative_fallback_raw_syntax_bad',
27
+ },
28
+ }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
32
 
 
 
 
 
33
 
34
  if __name__ == "__main__":
35
+ main()
persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0049/verify_workplace.py CHANGED
@@ -1,67 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
-
5
- def verify_workplace(workspace):
6
- score = 0
7
- details = []
8
-
9
- target_dir = os.path.join(workspace, "bug_report")
10
- target_file = os.path.join(target_dir, "culprit_symbol.txt")
11
-
12
- # 1. 检查目录和文件是否存在 (10分)
13
- if os.path.exists(target_dir) and os.path.exists(target_file):
14
- score += 10
15
- details.append({"item": "检查结果文件是否存在", "score": 10, "max_score": 10, "passed": True, "reason": "文件 bug_report/culprit_symbol.txt 已生成"})
16
- else:
17
- details.append({"item": "检查结果文件是否存在", "score": 0, "max_score": 10, "passed": False, "reason": "未找到 bug_report/culprit_symbol.txt"})
18
- # 如果文件不存在，后续检查无法进行
19
- save_score(score, details)
20
- return
21
 
22
- # 2. 检查文件内容是否正确 (80分)
23
- # 目标符号是 update_hardware_watchdog
24
- correct_symbol = "update_hardware_watchdog"
25
- try:
26
- with open(target_file, 'r', encoding='utf-8') as f:
27
- content = f.read().strip()
28
-
29
- if content == correct_symbol:
30
- score += 80
31
- details.append({"item": "检查符号名称是否匹配", "score": 80, "max_score": 80, "passed": True, "reason": f"成功定位到错误符号：{correct_symbol}"})
32
- elif correct_symbol in content and len(content) < 50:
33
- # 允许一些轻微的格式偏差，但扣分
34
- score += 40
35
- details.append({"item": "检查符号名称是否匹配", "score": 40, "max_score": 80, "passed": True, "reason": f"找到了正确符号但文件格式不符合‘纯文本’要求：{content}"})
36
- else:
37
- details.append({"item": "检查符号名称是否匹配", "score": 0, "max_score": 80, "passed": False, "reason": f"符号不匹配。预期：{correct_symbol}，实际：{content}"})
38
- except Exception as e:
39
- details.append({"item": "读取文件内容", "score": 0, "max_score": 80, "passed": False, "reason": f"读取失败: {str(e)}"})
40
 
41
- # 3. 检查文件整洁度 (10分)
42
- # 提示词要求“纯文本形式”，不应包含括号、参数或长篇大论
43
- try:
44
- with open(target_file, 'r', encoding='utf-8') as f:
45
- raw_content = f.read()
46
-
47
- if "(" in raw_content or ")" in raw_content or "\n" in raw_content.strip():
48
- details.append({"item": "检查输出格式整洁度", "score": 0, "max_score": 10, "passed": False, "reason": "文件中包含了多余的括号、换行或参数信息"})
49
- else:
50
- score += 10
51
- details.append({"item": "检查输出格式整洁度", "score": 10, "max_score": 10, "passed": True, "reason": "文件内容简洁，仅包含符号名"})
52
- except:
53
- pass
54
-
55
- save_score(score, details)
56
-
57
- def save_score(score, details):
58
- output = {
59
- "total_score": score,
60
- "details": details
61
  }
62
- with open("workplace_score.json", "w", encoding='utf-8') as f:
63
- json.dump(output, f, indent=2, ensure_ascii=False)
 
 
64
 
65
  if __name__ == "__main__":
66
- workspace = sys.argv[1] if len(sys.argv) > 1 else "."
67
- verify_workplace(workspace)
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
9
+ workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": "Original same-group verifier could not be materialized as an executable workplace verifier. Validation reasons: syntax_error:'break' outside loop:line_81.",
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'hard',
24
+ "source_task_id": 'data_92',
25
+ "imported_task_id": 'data_persona_aligned_hard_50_0049',
26
+ "action": 'conservative_fallback_raw_syntax_bad',
27
+ },
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
+ main()
 
persona_aligned_mix_200/tasks/data_persona_aligned_hard_50_0050/verify_workplace.py CHANGED
@@ -11,19 +11,19 @@ def main() -> None:
11
  "total_score": 0,
12
  "details": [
13
  {
14
- "item": "verifier_repair_fallback",
15
  "score": 0,
16
  "max_score": 100,
17
  "passed": False,
18
- "reason": 'Original ClawBenchPro verify_workplace.py could not be recovered from a same-source valid donor. Validation reasons: missing_score_output_marker; syntax_error:unterminated string literal (detected at line 72):line_72.',
19
  }
20
  ],
21
- "repair_metadata": {
22
  "dataset": 'persona_aligned_mix_200',
23
  "group": 'hard',
24
  "source_task_id": 'data_77',
25
  "imported_task_id": 'data_persona_aligned_hard_50_0050',
26
- "repair_action": "write_conservative_zero_score_fallback",
27
  },
28
  }
29
  output_path = os.path.join(workspace, "workplace_score.json")
 
11
  "total_score": 0,
12
  "details": [
13
  {
14
+ "item": "verifier_materialization_fallback",
15
  "score": 0,
16
  "max_score": 100,
17
  "passed": False,
18
+ "reason": 'Original same-group verifier could not be materialized as an executable workplace verifier. Validation reasons: syntax_error:unterminated string literal (detected at line 72):line_72.',
19
  }
20
  ],
21
+ "verifier_materialization": {
22
  "dataset": 'persona_aligned_mix_200',
23
  "group": 'hard',
24
  "source_task_id": 'data_77',
25
  "imported_task_id": 'data_persona_aligned_hard_50_0050',
26
+ "action": 'conservative_fallback_raw_syntax_bad',
27
  },
28
  }
29
  output_path = os.path.join(workspace, "workplace_score.json")
persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0001/verify_workplace.py CHANGED
@@ -1,68 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
-
5
- def verify():
6
- workspace = sys.argv[1] if len(sys.argv) > 1 else "."
7
- target_file = os.path.join(workspace, "triage", "conflict_target.json")
8
-
9
- score = 0
10
- details = []
11
 
12
- # 1. 检查文件是否存在与基础格式 (10分)
13
- if os.path.exists(target_file):
14
- try:
15
- with open(target_file, 'r', encoding='utf-8') as f:
16
- data = json.load(f)
17
- score += 10
18
- details.append({"item": "JSON文件存在且格式正确", "score": 10, "max_score": 10, "passed": True, "reason": "文件读取成功"})
19
- except Exception as e:
20
- details.append({"item": "JSON文件格式解析", "score": 0, "max_score": 10, "passed": False, "reason": f"解析失败: {str(e)}"})
21
- data = {}
22
- else:
23
- details.append({"item": "JSON文件是否存在", "score": 0, "max_score": 10, "passed": False, "reason": "未找到 triage/conflict_target.json"})
24
- data = {}
25
 
26
- # 预定义的标准答案 (根据 env_builder.py 的逻辑)
27
- # 冲突发生点：node-beta 在收到 node-gamma (T5) 的心跳时，本地 index 100 的 term 是 4
28
- expected_node = "node-beta"
29
- expected_term = 4
30
- expected_index = 100
31
-
32
- # 2. 检查 node_id (30分)
33
- node_id = data.get("node_id")
34
- if node_id == expected_node:
35
- score += 30
36
- details.append({"item": "匹配冲突节点 ID", "score": 30, "max_score": 30, "passed": True, "reason": f"成功识别节点: {node_id}"})
37
- else:
38
- details.append({"item": "匹配冲突节点 ID", "score": 0, "max_score": 30, "passed": False, "reason": f"期望 {expected_node}, 实际得到 {node_id}"})
39
-
40
- # 3. 检查 conflict_term (30分)
41
- try:
42
- term = int(data.get("conflict_term", -1))
43
- if term == expected_term:
44
- score += 30
45
- details.append({"item": "匹配冲突任期号 (Term)", "score": 30, "max_score": 30, "passed": True, "reason": f"成功识别任期: {term}"})
46
- else:
47
- details.append({"item": "匹配冲突任期号 (Term)", "score": 0, "max_score": 30, "passed": False, "reason": f"期望 {expected_term}, 实际得到 {term}"})
48
- except (ValueError, TypeError):
49
- details.append({"item": "匹配冲突任期号 (Term)", "score": 0, "max_score": 30, "passed": False, "reason": "任期号缺失或非整数"})
50
-
51
- # 4. 检查 conflict_index (30分)
52
- try:
53
- idx = int(data.get("conflict_index", -1))
54
- if idx == expected_index:
55
- score += 30
56
- details.append({"item": "匹配冲突日志索引 (Log Index)", "score": 30, "max_score": 30, "passed": True, "reason": f"成功识别索引: {idx}"})
57
- else:
58
- details.append({"item": "匹配冲突日志索引 (Log Index)", "score": 0, "max_score": 30, "passed": False, "reason": f"期望 {expected_index}, 实际得到 {idx}"})
59
- except (ValueError, TypeError):
60
- details.append({"item": "匹配冲突日志索引 (Log Index)", "score": 0, "max_score": 30, "passed": False, "reason": "索引号缺失或非整数"})
61
 
62
- # 结果写入
63
- output_file = "workplace_score.json"
64
- with open(output_file, 'w', encoding='utf-8') as f:
65
- json.dump({"total_score": score, "details": details}, f, indent=2, ensure_ascii=False)
66
 
67
  if __name__ == "__main__":
68
- verify()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
9
+ workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'This multi-turn task is evaluated from the original per-turn verifier scripts stored in verifiers/*.jsonl. The task-local verify_workplace.py is intentionally a non-authoritative zero-score fallback.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'multi_turn',
24
+ "source_task_id": 'data_94',
25
+ "imported_task_id": 'data_persona_aligned_multi_turn_50_0001',
26
+ "action": 'task_local_turn_verifier_placeholder',
27
+ },
28
+ }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
 
 
 
 
 
 
 
 
 
 
 
32
 
 
 
 
 
33
 
34
  if __name__ == "__main__":
35
+ main()
persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0002/verify_workplace.py CHANGED
@@ -1,78 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
-
5
- def verify_workplace():
6
- workspace = sys.argv[1] if len(sys.argv) > 1 else "."
7
- report_path = os.path.join(workspace, "ci_patch/conflict_report.json")
8
-
9
- score = 0
10
- details = []
11
-
12
- # 1. Check if the directory and file exist (10 points)
13
- if os.path.exists(report_path):
14
- score += 10
15
- details.append({"item": "检查结果文件是否存在", "score": 10, "max_score": 10, "passed": True, "reason": "文件 ci_patch/conflict_report.json 存在"})
16
-
17
- # 2. Check if the file is valid JSON (10 points)
18
- try:
19
- with open(report_path, "r", encoding="utf-8") as f:
20
- data = json.load(f)
21
- score += 10
22
- details.append({"item": "检查 JSON 格式合法性", "score": 10, "max_score": 10, "passed": True, "reason": "JSON 解析成功"})
23
-
24
- # 3. Check for required fields (10 points)
25
- required_fields = ["package", "version_a", "version_b"]
26
- missing_fields = [f for f in required_fields if f not in data]
27
- if not missing_fields:
28
- score += 10
29
- details.append({"item": "检查必填字段", "score": 10, "max_score": 10, "passed": True, "reason": "包含所有必需字段"})
30
-
31
- # 4. Check package name (30 points)
32
- # Allow case-insensitive check and trim
33
- actual_package = str(data.get("package", "")).strip().lower()
34
- expected_package = "eigen_matrix"
35
- if actual_package == expected_package:
36
- score += 30
37
- details.append({"item": "验证冲突包名", "score": 30, "max_score": 30, "passed": True, "reason": f"匹配正确: {expected_package}"})
38
- else:
39
- details.append({"item": "验证冲突包名", "score": 0, "max_score": 30, "passed": False, "reason": f"期望 {expected_package}, 实际为 {actual_package}"})
40
 
41
- # 5. Check versions (20 + 20 points)
42
- # Versions might be swapped, we accept both orders
43
- actual_versions = sorted([str(data.get("version_a", "")), str(data.get("version_b", ""))])
44
- expected_versions = sorted(["3.3.9", "3.4.2"])
45
-
46
- if actual_versions[0] == expected_versions[0]:
47
- score += 20
48
- details.append({"item": "验证版本号 A", "score": 20, "max_score": 20, "passed": True, "reason": f"版本 {actual_versions[0]} 匹配成功"})
49
- else:
50
- details.append({"item": "验证版本号 A", "score": 0, "max_score": 20, "passed": False, "reason": f"未找到版本 {expected_versions[0]}"})
51
 
52
- if actual_versions[1] == expected_versions[1]:
53
- score += 20
54
- details.append({"item": "验证版本号 B", "score": 20, "max_score": 20, "passed": True, "reason": f"版本 {actual_versions[1]} 匹配成功"})
55
- else:
56
- details.append({"item": "验证版本号 B", "score": 0, "max_score": 20, "passed": False, "reason": f"未找到版本 {expected_versions[1]}"})
57
-
58
- else:
59
- details.append({"item": "检查必填字段", "score": 0, "max_score": 10, "passed": False, "reason": f"缺失字段: {missing_fields}"})
60
- details.append({"item": "验证详细内容", "score": 0, "max_score": 70, "passed": False, "reason": "由于 JSON 字段不全，无法进行内容比对"})
61
-
62
- except json.JSONDecodeError:
63
- details.append({"item": "检查 JSON 格式合法性", "score": 0, "max_score": 10, "passed": False, "reason": "JSON 格式错误，无法解析"})
64
- details.append({"item": "验证后续内容", "score": 0, "max_score": 80, "passed": False, "reason": "由于 JSON 解析失败，跳过内容验证"})
65
- else:
66
- details.append({"item": "检查结果文件是否存在", "score": 0, "max_score": 10, "passed": False, "reason": "文件 ci_patch/conflict_report.json 未找到"})
67
- details.append({"item": "验证后续所有项", "score": 0, "max_score": 90, "passed": False, "reason": "找不到目标文件"})
68
-
69
- # Output results
70
- output_data = {
71
- "total_score": score,
72
- "details": details
73
  }
74
- with open("workplace_score.json", "w", encoding="utf-8") as f:
75
- json.dump(output_data, f, ensure_ascii=False, indent=2)
 
 
76
 
77
  if __name__ == "__main__":
78
- verify_workplace()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
9
+ workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'This multi-turn task is evaluated from the original per-turn verifier scripts stored in verifiers/*.jsonl. The task-local verify_workplace.py is intentionally a non-authoritative zero-score fallback.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'multi_turn',
24
+ "source_task_id": 'data_24',
25
+ "imported_task_id": 'data_persona_aligned_multi_turn_50_0002',
26
+ "action": 'task_local_turn_verifier_placeholder',
27
+ },
 
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
+ main()
persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0003/verify_workplace.py CHANGED
@@ -1,105 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
-
5
- def calculate_average_phred(qual_str):
6
- if not qual_str:
7
- return 0
8
- return sum(ord(c) - 33 for c in qual_str) / len(qual_str)
9
-
10
- def get_gold_standard(fastq_path):
11
- adapter = "GATCGGAAGAGCACACGTC"
12
- surviving_ids = []
13
-
14
- if not os.path.exists(fastq_path):
15
- return surviving_ids
16
 
17
- with open(fastq_path, 'r') as f:
18
- while True:
19
- header = f.readline().strip()
20
- if not header:
21
- break
22
- seq = f.readline().strip()
23
- plus = f.readline().strip()
24
- qual = f.readline().strip()
25
-
26
- # 规则 1: 剔除包含接头的
27
- if adapter in seq:
28
- continue
29
-
30
- # 规则 2: 平均质量分数低于 20 的剔除
31
- if calculate_average_phred(qual) < 20:
32
- continue
33
-
34
- # 提取 ID (去除开头的 @)
35
- read_id = header[1:] if header.startswith('@') else header
36
- surviving_ids.append(read_id)
37
-
38
- return surviving_ids
39
 
40
- def main():
41
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
42
- fastq_path = os.path.join(workspace, "raw_data/run_774.fastq")
43
- result_path = os.path.join(workspace, "results/surviving_reads.txt")
44
-
45
- score = 0
46
- details = []
47
-
48
- # 1. 检查结果文件是否存在
49
- if os.path.exists(result_path):
50
- score += 10
51
- details.append({"item": "结果文件存在性", "score": 10, "max_score": 10, "passed": True, "reason": "results/surviving_reads.txt 已生成"})
52
- else:
53
- details.append({"item": "结果文件存在性", "score": 0, "max_score": 10, "passed": False, "reason": "未找到 results/surviving_reads.txt"})
54
- # 如果文件不存在，后续检查无法进行
55
- with open("workplace_score.json", "w") as f:
56
- json.dump({"total_score": 0, "details": details}, f)
57
- return
58
-
59
- # 2. 读取并验证结果格式
60
- with open(result_path, 'r') as f:
61
- agent_lines = [line.strip() for line in f.readlines() if line.strip()]
62
-
63
- has_at_prefix = any(line.startswith('@') for line in agent_lines)
64
- if not has_at_prefix:
65
- score += 20
66
- details.append({"item": "输出格式正确性（无@前缀）", "score": 20, "max_score": 20, "passed": True, "reason": "Read ID 符合要求，没有包含 @ 符号"})
67
- else:
68
- details.append({"item": "输出格式正确性（无@前缀）", "score": 0, "max_score": 20, "passed": False, "reason": "部分 Read ID 仍保留了 FASTQ 的 @ 前缀"})
69
-
70
- # 3. 逻辑验证（金标准比对）
71
- gold_ids = set(get_gold_standard(fastq_path))
72
- agent_ids = set(agent_lines)
73
-
74
- # 计算交集、差集
75
- tp = len(gold_ids.intersection(agent_ids))
76
- fp = len(agent_ids - gold_ids)
77
- fn = len(gold_ids - agent_ids)
78
-
79
- if len(gold_ids) == 0:
80
- accuracy_score = 0 # 异常情况
81
- else:
82
- # 允许极小误差，但逻辑错误（如没过滤接头或质量分算错）会导致大量差异
83
- accuracy = tp / len(gold_ids) if len(gold_ids) > 0 else 0
84
- penalty = (fp / len(gold_ids)) * 0.5 # 错选惩罚
85
-
86
- final_acc_score = max(0, (accuracy - penalty) * 70)
87
- score += int(final_acc_score)
88
-
89
- if final_acc_score >= 65:
90
- details.append({"item": "数据过滤准确性", "score": int(final_acc_score), "max_score": 70, "passed": True, "reason": f"匹配率极高: 命中 {tp}, 多选 {fp}, 漏选 {fn}"})
91
- else:
92
- details.append({"item": "数据过滤准确性", "score": int(final_acc_score), "max_score": 70, "passed": False, "reason": f"过滤逻辑存在明显偏差: 命中 {tp}, 多选 {fp}, 漏选 {fn}"})
93
-
94
- # 总结输出
95
- final_score = min(100, score)
96
- output = {
97
- "total_score": int(final_score),
98
- "details": details
99
  }
100
-
101
- with open("workplace_score.json", "w") as f:
102
- json.dump(output, f, indent=2)
 
103
 
104
  if __name__ == "__main__":
105
  main()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'Original same-group verifier could not be materialized as an executable workplace verifier. Validation reasons: turn_1:empty; turn_1:missing_score_output_marker; turn_2:empty; turn_2:missing_score_output_marker; turn_3:empty; turn_3:missing_score_output_marker.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'multi_turn',
24
+ "source_task_id": 'data_06',
25
+ "imported_task_id": 'data_persona_aligned_multi_turn_50_0003',
26
+ "action": 'conservative_fallback_raw_empty',
27
+ },
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
  main()
persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0004/verify_workplace.py CHANGED
@@ -1,79 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
-
5
- def verify():
6
- workspace = sys.argv[1] if len(sys.argv) > 1 else "."
7
- score = 0
8
- details = []
9
-
10
- target_file = os.path.join(workspace, "calibration/ghost_ids.txt")
11
-
12
- # 1. 基础检查：文件是否存在 (10分)
13
- if os.path.exists(target_file):
14
- score += 10
15
- details.append({"item": "检查结果文件是否存在", "score": 10, "max_score": 10, "passed": True, "reason": "文件 ghost_ids.txt 存在"})
16
-
17
- try:
18
- with open(target_file, "r", encoding="utf-8") as f:
19
- content = f.read().strip()
20
-
21
- # 解析内容
22
- raw_ids = [s.strip() for s in content.split(",") if s.strip()]
23
- try:
24
- agent_ghost_ids = set(int(rid) for rid in raw_ids)
25
- except ValueError:
26
- details.append({"item": "检查文件内容格式", "score": 0, "max_score": 10, "passed": False, "reason": f"文件内容无法解析为整数列表: {content}"})
27
- return score, details
28
 
29
- # 2. 格式合规性检查 (10分)
30
- # 要求：仅逗号分隔，无额外描述
31
- if len(content.split("\n")) == 1 and "," in content:
32
- score += 10
33
- details.append({"item": "检查输出格式合规性", "score": 10, "max_score": 10, "passed": True, "reason": "格式符合逗号分隔要求"})
34
- else:
35
- details.append({"item": "检查输出格式合规性", "score": 0, "max_score": 10, "passed": False, "reason": "格式不符合单行逗号分隔要求"})
36
 
37
- # 3. 核心逻辑：检测幽灵障碍物 ID 的准确性
38
- # 根据 env_builder.py，正确答案是 {18, 27, 42, 68}
39
- # 正常 ID 是 {12, 33, 55}
40
- ground_truth_ghosts = {18, 27, 42, 68}
41
- ground_truth_normals = {12, 33, 55}
42
-
43
- # 正确识别的幽灵 (每个15分，共60分)
44
- for gid in ground_truth_ghosts:
45
- if gid in agent_ghost_ids:
46
- score += 15
47
- details.append({"item": f"检测幽灵 ID {gid}", "score": 15, "max_score": 15, "passed": True, "reason": "正确识别"})
48
- else:
49
- details.append({"item": f"检测幽灵 ID {gid}", "score": 0, "max_score": 15, "passed": False, "reason": "漏检"})
50
-
51
- # 错误识别的正常障碍物 (误报，每个扣10分，最低到0分)
52
- for nid in ground_truth_normals:
53
- if nid in agent_ghost_ids:
54
- score -= 10
55
- details.append({"item": f"误报检查 ID {nid}", "score": -10, "max_score": 0, "passed": False, "reason": f"将正常目标 {nid} 误判为幽灵"})
56
-
57
- # 检查是否包含虚构的 ID
58
- for aid in agent_ghost_ids:
59
- if aid not in ground_truth_ghosts and aid not in ground_truth_normals:
60
- score -= 5
61
- details.append({"item": f"虚构检测 ID {aid}", "score": -5, "max_score": 0, "passed": False, "reason": "结果中包含数据中不存在的 ID"})
62
-
63
- except Exception as e:
64
- details.append({"item": "解析文件异常", "score": 0, "max_score": 80, "passed": False, "reason": str(e)})
65
- else:
66
- details.append({"item": "检查结果文件是否存在", "score": 0, "max_score": 100, "passed": False, "reason": "找不到 calibration/ghost_ids.txt"})
67
 
68
- # 最终分修正
69
- score = max(0, min(100, score))
70
- return score, details
71
 
72
  if __name__ == "__main__":
73
- total_score, score_details = verify()
74
- output = {
75
- "total_score": total_score,
76
- "details": score_details
77
- }
78
- with open("workplace_score.json", "w", encoding="utf-8") as f:
79
- json.dump(output, f, indent=2, ensure_ascii=False)
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
9
+ workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'This multi-turn task is evaluated from the original per-turn verifier scripts stored in verifiers/*.jsonl. The task-local verify_workplace.py is intentionally a non-authoritative zero-score fallback.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'multi_turn',
24
+ "source_task_id": 'data_28',
25
+ "imported_task_id": 'data_persona_aligned_multi_turn_50_0004',
26
+ "action": 'task_local_turn_verifier_placeholder',
27
+ },
28
+ }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
 
 
 
 
 
 
32
 
 
 
 
33
 
34
  if __name__ == "__main__":
35
+ main()
 
 
 
 
 
 
persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0005/verify_workplace.py CHANGED
@@ -1,96 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
 
5
- def load_json(path):
6
- try:
7
- with open(path, 'r', encoding='utf-8') as f:
8
- return json.load(f)
9
- except Exception:
10
- return None
11
 
12
- def verify():
13
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
14
- report_path = os.path.join(workspace, "reports/termination_list.json")
15
-
16
- score = 0
17
- details = []
18
-
19
- # 1. 基础结构检查 (10分)
20
- if not os.path.exists(report_path):
21
- details.append({"item": "文件存在性", "score": 0, "max_score": 10, "passed": False, "reason": "报告文件 reports/termination_list.json 未生成"})
22
- else:
23
- data = load_json(report_path)
24
- if isinstance(data, list):
25
- score += 10
26
- details.append({"item": "JSON 格式检查", "score": 10, "max_score": 10, "passed": True, "reason": "报告为合法的 JSON 数组"})
27
- else:
28
- details.append({"item": "JSON 格式检查", "score": 0, "max_score": 10, "passed": False, "reason": "报告格式不是 JSON 数组"})
29
- data = []
30
-
31
- # 2. 核心数据解析与验证准备
32
- # 预期结果集合 (根据 env_builder 生成逻辑)
33
- # EBS: vol-0abcd111111111111 (ai-core -> alice), vol-0abcd222222222222 (data-eng -> charlie), vol-0abcd333333333333 (unknown-team -> unknown)
34
- # EC2: i-0ffff111111111111 (ai-research -> bob)
35
- expected_ebs = {"vol-0abcd111111111111", "vol-0abcd222222222222", "vol-0abcd333333333333"}
36
- expected_ec2 = {"i-0ffff111111111111"}
37
-
38
- owners_map = {
39
- "vol-0abcd111111111111": "alice.ai@mega-corp.local",
40
- "vol-0abcd222222222222": "charlie.data@mega-corp.local",
41
- "vol-0abcd333333333333": "unknown",
42
- "i-0ffff111111111111": "bob.research@mega-corp.local"
43
- }
44
-
45
- if data:
46
- # 3. EBS 资源识别 (30分)
47
- found_ebs = {item['resource_id'] for item in data if item.get('resource_type') == 'EBS'}
48
- ebs_correct = found_ebs == expected_ebs
49
- ebs_excess = found_ebs - expected_ebs # 检查是否有误杀 (如 vol-0abcd999999999999)
50
-
51
- ebs_score = 0
52
- if "vol-0abcd111111111111" in found_ebs: ebs_score += 10
53
- if "vol-0abcd222222222222" in found_ebs: ebs_score += 10
54
- if "vol-0abcd333333333333" in found_ebs: ebs_score += 10
55
- if "vol-0abcd999999999999" in found_ebs: ebs_score -= 10 # 误杀扣分
56
-
57
- ebs_score = max(0, ebs_score)
58
- score += ebs_score
59
- details.append({"item": "EBS 闲置资源识别", "score": ebs_score, "max_score": 30, "passed": ebs_score >= 30, "reason": f"识别到 EBS: {found_ebs}"})
60
-
61
- # 4. EC2 资源识别 (30分)
62
- found_ec2 = {item['resource_id'] for item in data if item.get('resource_type') == 'EC2'}
63
- ec2_score = 0
64
- if "i-0ffff111111111111" in found_ec2: ec2_score += 30
65
- if "i-0ffff222222222222" in found_ec2 or "i-0ffff333333333333" in found_ec2:
66
- ec2_score = max(0, ec2_score - 20) # 误将高负载或平均负载正常的实例列入
67
-
68
- score += ec2_score
69
- details.append({"item": "EC2 GPU低利用率识别", "score": ec2_score, "max_score": 30, "passed": ec2_score == 30, "reason": f"识别到 EC2: {found_ec2}"})
70
-
71
- # 5. Owner 匹配正确性 (30分)
72
- owner_correct_count = 0
73
- total_items = len(data)
74
- for item in data:
75
- rid = item.get('resource_id')
76
- provided_owner = item.get('owner')
77
- if rid in owners_map and provided_owner == owners_map[rid]:
78
- owner_correct_count += 1
79
-
80
- owner_score = int((owner_correct_count / total_items * 30)) if total_items > 0 else 0
81
- score += owner_score
82
- details.append({"item": "负责人邮箱交叉比对", "score": owner_score, "max_score": 30, "passed": owner_score == 30, "reason": f"正确匹配了 {owner_correct_count}/{total_items} 个负责人"})
83
-
84
- # 最终分处理
85
- final_score = min(100, max(0, score))
86
-
87
  result = {
88
- "total_score": final_score,
89
- "details": details
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
90
  }
91
-
92
- with open("workplace_score.json", "w", encoding="utf-8") as f:
93
- json.dump(result, f, indent=2, ensure_ascii=False)
 
94
 
95
  if __name__ == "__main__":
96
- verify()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
6
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
10
  result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'This multi-turn task is evaluated from the original per-turn verifier scripts stored in verifiers/*.jsonl. The task-local verify_workplace.py is intentionally a non-authoritative zero-score fallback.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'multi_turn',
24
+ "source_task_id": 'data_38',
25
+ "imported_task_id": 'data_persona_aligned_multi_turn_50_0005',
26
+ "action": 'task_local_turn_verifier_placeholder',
27
+ },
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
+ main()
persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0006/verify_workplace.py CHANGED
@@ -1,120 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import httpx
5
- from openai import OpenAI
6
 
7
- # Configuration for potential LLM usage
8
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
9
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
10
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
11
 
12
- http_client = httpx.Client(verify=False)
13
- client = OpenAI(
14
- api_key=MOCK_API_KEY,
15
- base_url=MOCK_API_BASE,
16
- http_client=http_client
17
- )
18
-
19
- def llm_judge_content(prompt_text, file_content):
20
- try:
21
- response = client.chat.completions.create(
22
- model=MOCK_MODEL_NAME,
23
- messages=[
24
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
25
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
26
- ],
27
- temperature=0
28
- )
29
- return "yes" in response.choices[0].message.content.strip().lower()
30
- except Exception as e:
31
- print(f"LLM API Error: {e}")
32
- return False
33
-
34
- def verify():
35
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
36
- score = 0
37
- details = []
38
-
39
- output_path = os.path.join(workspace, "analysis/valid_p300_peaks.json")
40
-
41
- # 1. Check file existence (10 points)
42
- if os.path.exists(output_path):
43
- score += 10
44
- details.append({"item": "Output file existence", "score": 10, "max_score": 10, "passed": True, "reason": "Found analysis/valid_p300_peaks.json"})
45
-
46
- # 2. JSON Validity & Structure (10 points)
47
- try:
48
- with open(output_path, 'r', encoding='utf-8') as f:
49
- content = f.read()
50
- data = json.loads(content)
51
- score += 10
52
- details.append({"item": "JSON validity", "score": 10, "max_score": 10, "passed": True, "reason": "File is valid JSON"})
53
-
54
- # 3. Precision Check: EVT_001 (25 points)
55
- # Expected: 14.5
56
- if "EVT_001" in data and abs(float(data["EVT_001"]) - 14.5) < 0.01:
57
- score += 25
58
- details.append({"item": "EVT_001 Correctness", "score": 25, "max_score": 25, "passed": True, "reason": "Correct peak (14.5) for EVT_001"})
59
- else:
60
- details.append({"item": "EVT_001 Correctness", "score": 0, "max_score": 25, "passed": False, "reason": f"Expected 14.5, got {data.get('EVT_001')}"})
61
-
62
- # 4. Precision Check: EVT_005 (25 points)
63
- # Expected: 18.2
64
- if "EVT_005" in data and abs(float(data["EVT_005"]) - 18.2) < 0.01:
65
- score += 25
66
- details.append({"item": "EVT_005 Correctness", "score": 25, "max_score": 25, "passed": True, "reason": "Correct peak (18.2) for EVT_005"})
67
- else:
68
- details.append({"item": "EVT_005 Correctness", "score": 0, "max_score": 25, "passed": False, "reason": f"Expected 18.2, got {data.get('EVT_005')}"})
69
-
70
- # 5. Artifact Rejection: EVT_003 and EVT_004 (20 points total)
71
- # EVT_003 has FZ artifact, EVT_004 has CZ artifact.
72
- rejected_003 = "EVT_003" not in data
73
- rejected_004 = "EVT_004" not in data
74
-
75
- if rejected_003:
76
- score += 10
77
- details.append({"item": "Artifact Rejection (FZ)", "score": 10, "max_score": 10, "passed": True, "reason": "Correctly rejected EVT_003 due to FZ spike"})
78
- else:
79
- details.append({"item": "Artifact Rejection (FZ)", "score": 0, "max_score": 10, "passed": False, "reason": "Failed to reject EVT_003 (FZ artifact)"})
80
-
81
- if rejected_004:
82
- score += 10
83
- details.append({"item": "Artifact Rejection (CZ)", "score": 10, "max_score": 10, "passed": True, "reason": "Correctly rejected EVT_004 due to CZ spike"})
84
- else:
85
- details.append({"item": "Artifact Rejection (CZ)", "score": 0, "max_score": 10, "passed": False, "reason": "Failed to reject EVT_004 (CZ artifact)"})
86
-
87
- # 6. Type Filtering: EVT_002 (10 points)
88
- # EVT_002 is N200, should be ignored.
89
- if "EVT_002" not in data:
90
- score += 10
91
- details.append({"item": "Target Type Filtering", "score": 10, "max_score": 10, "passed": True, "reason": "Correctly ignored non-P300 stimulus EVT_002"})
92
- else:
93
- details.append({"item": "Target Type Filtering", "score": 0, "max_score": 10, "passed": False, "reason": "Failed to filter out non-P300 stimulus"})
94
-
95
- # Bonus/Cleanup: No extra verbosity check (LLM)
96
- # The prompt requested NO code explanations in the output.
97
- is_clean = llm_judge_content("Does the provided JSON file contain ONLY the stimulus-to-peak-voltage mapping without any conversational filler, explanations, or code commentary?", content)
98
- if not is_clean:
99
- penalty = 10
100
- score = max(0, score - penalty)
101
- details.append({"item": "Output Cleanliness", "score": -penalty, "max_score": 0, "passed": False, "reason": "Output contained forbidden explanations or commentary"})
102
-
103
- except Exception as e:
104
- details.append({"item": "JSON Parsing", "score": 0, "max_score": 10, "passed": False, "reason": f"Error parsing JSON: {str(e)}"})
105
- else:
106
- details.append({"item": "Output file existence", "score": 0, "max_score": 100, "passed": False, "reason": "analysis/valid_p300_peaks.json not found"})
107
-
108
- # Ensure score is integer and capped
109
- final_score = min(100, max(0, int(score)))
110
-
111
  result = {
112
- "total_score": final_score,
113
- "details": details
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
114
  }
115
-
116
- with open("workplace_score.json", "w", encoding="utf-8") as f:
117
- json.dump(result, f, indent=2, ensure_ascii=False)
 
118
 
119
  if __name__ == "__main__":
120
- verify()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
6
 
 
 
 
 
7
 
8
+ def main() -> None:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
10
  result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'This multi-turn task is evaluated from the original per-turn verifier scripts stored in verifiers/*.jsonl. The task-local verify_workplace.py is intentionally a non-authoritative zero-score fallback.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'multi_turn',
24
+ "source_task_id": 'data_75',
25
+ "imported_task_id": 'data_persona_aligned_multi_turn_50_0006',
26
+ "action": 'task_local_turn_verifier_placeholder',
27
+ },
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
+ main()
persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0007/verify_workplace.py CHANGED
@@ -1,116 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import math
5
 
6
- def calculate_expected_norm(fx, fy, fz):
7
- return math.sqrt(fx**2 + fy**2 + fz**2)
8
 
9
- def verify():
10
- # 默认工作区路径获取
11
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
12
- report_path = os.path.join(workspace, "report/culprit.json")
13
- score_file = "workplace_score.json"
14
-
15
- total_score = 0
16
- details = []
17
-
18
- # 预设的正确答案（基于 env_builder.py）
19
- EXPECTED_STEP = 14
20
- EXPECTED_ATOM_IDX = 42
21
- # fx, fy, fz = 845.210, -991.330, 1502.440
22
- EXPECTED_FORCE_NORM = calculate_expected_norm(845.210, -991.330, 1502.440) # 约 1988.577...
23
-
24
- # 1. 检查目录和文件是否存在 (10分)
25
- if os.path.exists(report_path):
26
- score_1 = 10
27
- details.append({"item": "检查报告文件是否存在", "score": 10, "max_score": 10, "passed": True, "reason": "找到 report/culprit.json"})
28
- else:
29
- score_1 = 0
30
- details.append({"item": "检查报告文件是否存在", "score": 0, "max_score": 10, "passed": False, "reason": "未找到 report/culprit.json"})
31
- # 如果文件不存在，后续检查无法进行，直接写入结果
32
- with open(score_file, "w") as f:
33
- json.dump({"total_score": 0, "details": details}, f, indent=2)
34
- return
35
-
36
- # 2. 检查 JSON 格式合法性 (10分)
37
- try:
38
- with open(report_path, 'r', encoding='utf-8') as f:
39
- data = json.load(f)
40
- score_2 = 10
41
- details.append({"item": "检查 JSON 格式合法性", "score": 10, "max_score": 10, "passed": True, "reason": "JSON 解析成功"})
42
- except Exception as e:
43
- score_2 = 0
44
- details.append({"item": "检查 JSON 格式合法性", "score": 0, "max_score": 10, "passed": False, "reason": f"JSON 解析失败: {str(e)}"})
45
- with open(score_file, "w") as f:
46
- json.dump({"total_score": score_1, "details": details}, f, indent=2)
47
- return
48
-
49
- # 3. 验证离子步序号 (20分)
50
- # 字段名可能不唯一，允许 agent 使用常用字段名，但优先匹配题目要求的逻辑
51
- step_keys = ["ionic_step", "step", "step_number", "fatal_step"]
52
- found_step = None
53
- for k in step_keys:
54
- if k in data:
55
- found_step = data[k]
56
- break
57
-
58
- if found_step == EXPECTED_STEP:
59
- score_3 = 20
60
- details.append({"item": "验证致命离子步序号", "score": 20, "max_score": 20, "passed": True, "reason": f"离子步序号正确: {found_step}"})
61
- else:
62
- score_3 = 0
63
- details.append({"item": "验证致命离子步序号", "score": 0, "max_score": 20, "passed": False, "reason": f"序号错误或缺失，期望 {EXPECTED_STEP}，实际拿到 {found_step}"})
64
-
65
- # 4. 验证原子索引 (30分)
66
- atom_keys = ["atom_index", "culprit_atom", "atom_id", "atom_idx"]
67
- found_atom = None
68
- for k in atom_keys:
69
- if k in data:
70
- found_atom = data[k]
71
- break
72
-
73
- if found_atom == EXPECTED_ATOM_IDX:
74
- score_4 = 30
75
- details.append({"item": "验证异常原子索引", "score": 30, "max_score": 30, "passed": True, "reason": f"原子索引正确: {found_atom}"})
76
- else:
77
- score_4 = 0
78
- details.append({"item": "验证异常原子索引", "score": 0, "max_score": 30, "passed": False, "reason": f"索引错误或缺失，期望 {EXPECTED_ATOM_IDX}，实际拿到 {found_atom}"})
79
-
80
- # 5. 验证受力大小 (30分)
81
- force_keys = ["force_magnitude", "force_norm", "max_force", "force"]
82
- found_force = None
83
- for k in force_keys:
84
- if k in data:
85
- found_force = data[k]
86
- break
87
-
88
- if found_force is not None:
89
- try:
90
- val = float(found_force)
91
- if math.isclose(val, EXPECTED_FORCE_NORM, rel_tol=1e-3):
92
- score_5 = 30
93
- details.append({"item": "验证受力绝对值计算", "score": 30, "max_score": 30, "passed": True, "reason": f"受力大小符合预期: {val}"})
94
- else:
95
- score_5 = 0
96
- details.append({"item": "验证受力绝对值计算", "score": 0, "max_score": 30, "passed": False, "reason": f"数值偏差过大，期望约 {EXPECTED_FORCE_NORM:.4f}, 实际为 {val}"})
97
- except:
98
- score_5 = 0
99
- details.append({"item": "验证受力绝对值计算", "score": 0, "max_score": 30, "passed": False, "reason": "受力字段无法转换为浮点数"})
100
- else:
101
- score_5 = 0
102
- details.append({"item": "验证受力绝对值计算", "score": 0, "max_score": 30, "passed": False, "reason": "未找到受力大小字段"})
103
-
104
- # 汇总
105
- total_score = score_1 + score_2 + score_3 + score_4 + score_5
106
-
107
- # 额外检查：如果 Agent 提供了多余的虚假字段（如猜测的化学元素等题目没给的信息），酌情扣分 (可选防御性逻辑)
108
- if len(data) > 6:
109
- total_score = max(0, total_score - 5)
110
- details.append({"item": "冗余信息惩罚", "score": -5, "max_score": 0, "passed": False, "reason": "JSON中包含过量未要求的字段，可能存在幻觉"})
111
 
112
- with open(score_file, "w") as f:
113
- json.dump({"total_score": int(total_score), "details": details}, f, indent=2)
114
 
115
  if __name__ == "__main__":
116
- verify()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
6
 
 
 
7
 
8
+ def main() -> None:
 
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'This multi-turn task is evaluated from the original per-turn verifier scripts stored in verifiers/*.jsonl. The task-local verify_workplace.py is intentionally a non-authoritative zero-score fallback.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'multi_turn',
24
+ "source_task_id": 'data_15',
25
+ "imported_task_id": 'data_persona_aligned_multi_turn_50_0007',
26
+ "action": 'task_local_turn_verifier_placeholder',
27
+ },
28
+ }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
32
 
 
 
33
 
34
  if __name__ == "__main__":
35
+ main()
persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0008/verify_workplace.py CHANGED
@@ -1,61 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
-
5
- def verify():
6
- # 基础路径处理
7
- workspace = sys.argv[1] if len(sys.argv) > 1 else "."
8
- report_path = os.path.join(workspace, "reports/bottleneck.json")
9
-
10
- score = 0
11
- details = []
12
-
13
- # 1. 检查结果文件是否存在 (10分)
14
- if os.path.exists(report_path):
15
- score += 10
16
- details.append({"item": "Check reports/bottleneck.json existence", "score": 10, "max_score": 10, "passed": True, "reason": "Report file found."})
17
- else:
18
- details.append({"item": "Check reports/bottleneck.json existence", "score": 0, "max_score": 10, "passed": False, "reason": "Report file not found."})
19
- # 如果文件不存在，后续检查无法进行，直接写入结果
20
- write_score(score, details)
21
- return
22
 
23
- # 2. 检查 JSON 格式与 Schema 合法性 (20分)
24
- try:
25
- with open(report_path, 'r', encoding='utf-8') as f:
26
- data = json.load(f)
27
-
28
- if "bottleneck_entity" in data:
29
- score += 20
30
- details.append({"item": "Verify JSON Schema", "score": 20, "max_score": 20, "passed": True, "reason": "Key 'bottleneck_entity' exists."})
31
-
32
- # 3. 检查 Entity ID 准确性 (70分)
33
- # 根据 env_builder.py, target_entity = "8847291"
34
- actual_id = str(data["bottleneck_entity"]).strip()
35
- expected_id = "8847291"
36
-
37
- if actual_id == expected_id:
38
- score += 70
39
- details.append({"item": "Verify Bottleneck Entity ID", "score": 70, "max_score": 70, "passed": True, "reason": f"Correct entity ID found: {actual_id}"})
40
- else:
41
- details.append({"item": "Verify Bottleneck Entity ID", "score": 0, "max_score": 70, "passed": False, "reason": f"Incorrect entity ID. Expected {expected_id}, but got {actual_id}"})
42
-
43
- else:
44
- details.append({"item": "Verify JSON Schema", "score": 0, "max_score": 20, "passed": False, "reason": "Required key 'bottleneck_entity' is missing."})
45
 
46
- except json.JSONDecodeError:
47
- details.append({"item": "Verify JSON Schema", "score": 0, "max_score": 20, "passed": False, "reason": "File is not a valid JSON."})
48
- except Exception as e:
49
- details.append({"item": "Verify JSON Schema", "score": 0, "max_score": 20, "passed": False, "reason": f"Unexpected error: {str(e)}"})
50
-
51
- write_score(score, details)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
52
 
53
- def write_score(total_score, details):
54
- with open("workplace_score.json", "w", encoding="utf-8") as f:
55
- json.dump({
56
- "total_score": total_score,
57
- "details": details
58
- }, f, indent=2, ensure_ascii=False)
59
 
60
  if __name__ == "__main__":
61
- verify()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
9
+ workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'This multi-turn task is evaluated from the original per-turn verifier scripts stored in verifiers/*.jsonl. The task-local verify_workplace.py is intentionally a non-authoritative zero-score fallback.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'multi_turn',
24
+ "source_task_id": 'data_61',
25
+ "imported_task_id": 'data_persona_aligned_multi_turn_50_0008',
26
+ "action": 'task_local_turn_verifier_placeholder',
27
+ },
28
+ }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
 
 
 
 
 
 
 
33
 
34
  if __name__ == "__main__":
35
+ main()
persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0009/verify_workplace.py CHANGED
@@ -1,216 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import math
5
- import httpx
6
- from openai import OpenAI
7
-
8
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
9
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
10
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
11
-
12
- # 初始化客户端，强制关闭 SSL 验证
13
- http_client = httpx.Client(verify=False)
14
- client = OpenAI(
15
- api_key=MOCK_API_KEY,
16
- base_url=MOCK_API_BASE,
17
- http_client=http_client
18
- )
19
-
20
- def llm_judge_content(prompt_text, file_content):
21
- try:
22
- response = client.chat.completions.create(
23
- model=MOCK_MODEL_NAME,
24
- messages=[
25
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
26
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
27
- ],
28
- temperature=0
29
- )
30
- return "yes" in response.choices[0].message.content.strip().lower()
31
- except Exception as e:
32
- print(f"LLM API Error: {e}")
33
- return False
34
-
35
- def verify(workspace):
36
- score_details = []
37
- total_score = 0
38
- target_file = os.path.join(workspace, "output", "critical_state.json")
39
-
40
- # 1. 检查文件是否存在 (20分)
41
- if not os.path.exists(target_file):
42
- score_details.append({
43
- "item": "检查结果文件是否存在",
44
- "score": 0,
45
- "max_score": 20,
46
- "passed": False,
47
- "reason": f"未找到文件 {target_file}"
48
- })
49
- write_score(0, score_details, workspace)
50
- return
51
- else:
52
- score_details.append({
53
- "item": "检查结果文件是否存在",
54
- "score": 20,
55
- "max_score": 20,
56
- "passed": True,
57
- "reason": "文件 output/critical_state.json 存在"
58
- })
59
- total_score += 20
60
-
61
- # 2. 检查 JSON 格式合法性 (15分)
62
- try:
63
- with open(target_file, "r", encoding="utf-8") as f:
64
- data = json.load(f)
65
- score_details.append({
66
- "item": "检查 JSON 解析",
67
- "score": 15,
68
- "max_score": 15,
69
- "passed": True,
70
- "reason": "文件为合法 JSON"
71
- })
72
- total_score += 15
73
- except Exception as e:
74
- score_details.append({
75
- "item": "检查 JSON 解析",
76
- "score": 0,
77
- "max_score": 15,
78
- "passed": False,
79
- "reason": f"解析 JSON 失败: {e}"
80
- })
81
- write_score(total_score, score_details, workspace)
82
- return
83
 
84
- # 3. 检查 JSON 键名准确性与无幻觉字段 (15分)
85
- expected_keys = {"latest_quaternion", "max_temperature"}
86
- actual_keys = set(data.keys())
87
- if actual_keys == expected_keys:
88
- score_details.append({
89
- "item": "检查 JSON 字段严格匹配",
90
- "score": 15,
91
- "max_score": 15,
92
- "passed": True,
93
- "reason": "字段名称完全匹配，无多余捏造字段"
94
- })
95
- total_score += 15
96
- else:
97
- missing = expected_keys - actual_keys
98
- extra = actual_keys - expected_keys
99
- reason_parts = []
100
- if missing: reason_parts.append(f"缺失: {missing}")
101
- if extra: reason_parts.append(f"多余: {extra}")
102
- score_details.append({
103
- "item": "检查 JSON 字段严格匹配",
104
- "score": 0,
105
- "max_score": 15,
106
- "passed": False,
107
- "reason": "字段不完全匹配。 " + " | ".join(reason_parts)
108
- })
109
 
110
- # 4. 检查 max_temperature 计算结果 (25分)
111
- temp = data.get("max_temperature", None)
112
- if temp is not None:
113
- try:
114
- temp_val = float(temp)
115
- # 正确值为 94.75。容忍度很低
116
- if math.isclose(temp_val, 94.75, abs_tol=0.01):
117
- score_details.append({
118
- "item": "验证最大异常温度峰值",
119
- "score": 25,
120
- "max_score": 25,
121
- "passed": True,
122
- "reason": "最高异常温度峰值精确等于 94.75"
123
- })
124
- total_score += 25
125
- else:
126
- score_details.append({
127
- "item": "验证最大异常温度峰值",
128
- "score": 0,
129
- "max_score": 25,
130
- "passed": False,
131
- "reason": f"温度值错误，期望 94.75，实际为 {temp_val}"
132
- })
133
- except ValueError:
134
- score_details.append({
135
- "item": "验证最大异常温度峰值",
136
- "score": 0,
137
- "max_score": 25,
138
- "passed": False,
139
- "reason": "max_temperature 并非有效数值类型"
140
- })
141
- else:
142
- score_details.append({
143
- "item": "验证最大异常温度峰值",
144
- "score": 0,
145
- "max_score": 25,
146
- "passed": False,
147
- "reason": "未找到 max_temperature 字段"
148
- })
149
-
150
- # 5. 检查 latest_quaternion 提取与计算结果 (25分)
151
- quat = data.get("latest_quaternion", None)
152
- if quat is not None:
153
- if isinstance(quat, list) and len(quat) == 4:
154
- expected_quat = [0.4999, 0.5001, -0.4999, -0.5001]
155
- try:
156
- match_all = True
157
- for val, exp in zip(quat, expected_quat):
158
- if not math.isclose(float(val), exp, abs_tol=0.0002):
159
- match_all = False
160
- break
161
- if match_all:
162
- score_details.append({
163
- "item": "验证最新星象仪四元数",
164
- "score": 25,
165
- "max_score": 25,
166
- "passed": True,
167
- "reason": f"成功提取有效时间最新的一帧四元数并保留正确小数位"
168
- })
169
- total_score += 25
170
- else:
171
- score_details.append({
172
- "item": "验证最新星象仪四元数",
173
- "score": 0,
174
- "max_score": 25,
175
- "passed": False,
176
- "reason": f"四元数值不匹配，可能找错了时间帧、提取到了被破坏的帧头数据或解析小/大端序出错。实际值：{quat}"
177
- })
178
- except ValueError:
179
- score_details.append({
180
- "item": "验证最新星象仪四元数",
181
- "score": 0,
182
- "max_score": 25,
183
- "passed": False,
184
- "reason": "数组内含有非数值数据"
185
- })
186
- else:
187
- score_details.append({
188
- "item": "验证最新星象仪四元数",
189
  "score": 0,
190
- "max_score": 25,
191
  "passed": False,
192
- "reason": "latest_quaternion 格式错误，必须为包含4个数值的数组"
193
- })
194
- else:
195
- score_details.append({
196
- "item": "验证最新星象仪四元数",
197
- "score": 0,
198
- "max_score": 25,
199
- "passed": False,
200
- "reason": "未找到 latest_quaternion 字段"
201
- })
202
-
203
- write_score(total_score, score_details, workspace)
204
-
205
- def write_score(total_score, details, workspace):
206
- result = {
207
- "total_score": total_score,
208
- "details": details
209
  }
210
- with open(os.path.join(workspace, "workplace_score.json"), "w", encoding="utf-8") as f:
211
- json.dump(result, f, indent=2, ensure_ascii=False)
212
- print(json.dumps(result, indent=2, ensure_ascii=False))
 
213
 
214
  if __name__ == "__main__":
215
- work_dir = sys.argv[1] if len(sys.argv) > 1 else "."
216
- verify(work_dir)
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
9
+ workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
15
  "score": 0,
16
+ "max_score": 100,
17
  "passed": False,
18
+ "reason": 'This multi-turn task is evaluated from the original per-turn verifier scripts stored in verifiers/*.jsonl. The task-local verify_workplace.py is intentionally a non-authoritative zero-score fallback.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'multi_turn',
24
+ "source_task_id": 'data_58',
25
+ "imported_task_id": 'data_persona_aligned_multi_turn_50_0009',
26
+ "action": 'task_local_turn_verifier_placeholder',
27
+ },
 
 
 
 
 
 
 
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
+ main()
 
persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0010/verify_workplace.py CHANGED
@@ -1,117 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import httpx
5
- from openai import OpenAI
6
-
7
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
8
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
9
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
10
 
11
- # 初始化客户端，必须关闭 SSL 验证
12
- http_client = httpx.Client(verify=False)
13
- client = OpenAI(
14
- api_key=MOCK_API_KEY,
15
- base_url=MOCK_API_BASE,
16
- http_client=http_client
17
- )
18
 
19
- def llm_judge_content(prompt_text, file_content):
20
- # 此函数为检测非结构化文本的统一接口
21
- try:
22
- response = client.chat.completions.create(
23
- model=MOCK_MODEL_NAME,
24
- messages=[
25
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
26
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
27
- ],
28
- temperature=0
29
- )
30
- return "yes" in response.choices[0].message.content.strip().lower()
31
- except Exception as e:
32
- print(f"LLM API Error: {e}")
33
- return False
34
-
35
- def verify():
36
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
37
- target_file = os.path.join(workspace, "report", "failed_init.json")
38
-
39
- total_score = 0
40
- details = []
41
-
42
- # 1. 检查目标文件是否存在 (20分)
43
- if os.path.isfile(target_file):
44
- total_score += 20
45
- details.append({"item": "检查结果文件是否存在", "score": 20, "max_score": 20, "passed": True, "reason": "文件 report/failed_init.json 存在"})
46
- else:
47
- details.append({"item": "检查结果文件是否存在", "score": 0, "max_score": 20, "passed": False, "reason": "文件 report/failed_init.json 不存在"})
48
- write_score(total_score, details)
49
- return
50
-
51
- # 2. 检查文件是否为合法的 JSON 格式 (20分)
52
- try:
53
- with open(target_file, 'r', encoding='utf-8') as f:
54
- data = json.load(f)
55
- total_score += 20
56
- details.append({"item": "检查文件是否为合法 JSON", "score": 20, "max_score": 20, "passed": True, "reason": "成功解析 JSON 文件"})
57
- except json.JSONDecodeError:
58
- details.append({"item": "检查文件是否为合法 JSON", "score": 0, "max_score": 20, "passed": False, "reason": "文件内容不是合法的 JSON 格式"})
59
- write_score(total_score, details)
60
- return
61
- except Exception as e:
62
- details.append({"item": "检查文件是否为合法 JSON", "score": 0, "max_score": 20, "passed": False, "reason": f"文件读取发生未知错误: {str(e)}"})
63
- write_score(total_score, details)
64
- return
65
-
66
- # 3. 检查 JSON 字段完整性 (10分)
67
- if not isinstance(data, dict):
68
- details.append({"item": "检查 JSON 结构类型", "score": 0, "max_score": 10, "passed": False, "reason": "JSON 根节点必须是一个对象 (dict)"})
69
- write_score(total_score, details)
70
- return
71
-
72
- has_register = "register" in data
73
- has_value = "value" in data
74
- extra_keys = set(data.keys()) - {"register", "value"}
75
-
76
- if has_register and has_value:
77
- if extra_keys:
78
- # 存在冗余字段，扣5分
79
- total_score += 5
80
- details.append({"item": "检查 JSON 字段", "score": 5, "max_score": 10, "passed": False, "reason": f"包含了必要的字段，但存在冗余字段: {extra_keys}"})
81
- else:
82
- total_score += 10
83
- details.append({"item": "检查 JSON 字段", "score": 10, "max_score": 10, "passed": True, "reason": "精确包含了 register 和 value 字段"})
84
- else:
85
- details.append({"item": "检查 JSON 字段", "score": 0, "max_score": 10, "passed": False, "reason": "缺失 register 或 value 字段"})
86
- write_score(total_score, details)
87
- return
88
-
89
- # 4. 提取并校验 register 的值 (25分)
90
- reg_val = str(data.get("register", "")).strip().lower()
91
- if reg_val == "0x2a":
92
- total_score += 25
93
- details.append({"item": "校验报错的寄存器地址", "score": 25, "max_score": 25, "passed": True, "reason": "找到了正确的寄存器地址 0x2A"})
94
- else:
95
- details.append({"item": "校验报错的寄存器地址", "score": 0, "max_score": 25, "passed": False, "reason": f"寄存器地址错误或格式不标准，预期 '0x2A'，实际获取到 '{reg_val}'"})
96
-
97
- # 5. 提取并校验 value 的值 (25分)
98
- val_val = str(data.get("value", "")).strip().lower()
99
- if val_val == "0x7f":
100
- total_score += 25
101
- details.append({"item": "校验试图写入的错误数据", "score": 25, "max_score": 25, "passed": True, "reason": "找到了正确的报错数据 0x7F"})
102
- else:
103
- details.append({"item": "校验试图写入的错误数据", "score": 0, "max_score": 25, "passed": False, "reason": f"试图写入的数据错误或格式不标准，预期 '0x7F'，实际获取到 '{val_val}'"})
104
-
105
- write_score(total_score, details)
106
-
107
- def write_score(total_score, details):
108
- report = {
109
- "total_score": total_score,
110
- "details": details
111
  }
112
- with open("workplace_score.json", "w", encoding="utf-8") as f:
113
- json.dump(report, f, indent=2, ensure_ascii=False)
114
- print(json.dumps(report, indent=2, ensure_ascii=False))
 
115
 
116
  if __name__ == "__main__":
117
- verify()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'This multi-turn task is evaluated from the original per-turn verifier scripts stored in verifiers/*.jsonl. The task-local verify_workplace.py is intentionally a non-authoritative zero-score fallback.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'multi_turn',
24
+ "source_task_id": 'data_59',
25
+ "imported_task_id": 'data_persona_aligned_multi_turn_50_0010',
26
+ "action": 'task_local_turn_verifier_placeholder',
27
+ },
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
+ main()
persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0011/verify_workplace.py CHANGED
@@ -1,74 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
-
5
- def verify():
6
- workspace = sys.argv[1] if len(sys.argv) > 1 else "."
7
- score = 0
8
- details = []
9
-
10
- target_file = os.path.join(workspace, "ops/kill_target.json")
11
-
12
- # 预期结果
13
- EXPECTED_PID = 14920
14
- EXPECTED_XID = 9948271
15
 
16
- # 1. 检查文件是否存在 (10分)
17
- if os.path.exists(target_file):
18
- score += 10
19
- details.append({"item": "文件检查", "score": 10, "max_score": 10, "passed": True, "reason": "ops/kill_target.json 存在"})
20
-
21
- # 2. 检查 JSON 格式与合法性 (20分)
22
- try:
23
- with open(target_file, 'r', encoding='utf-8') as f:
24
- data = json.load(f)
25
-
26
- score += 20
27
- details.append({"item": "JSON格式验证", "score": 20, "max_score": 20, "passed": True, "reason": "JSON 解析成功"})
28
-
29
- # 3. 检查 PID 是否正确 (30分)
30
- actual_pid = data.get("pid")
31
- if actual_pid == EXPECTED_PID:
32
- score += 30
33
- details.append({"item": "PID 识别", "score": 30, "max_score": 30, "passed": True, "reason": "成功识别罪魁祸首 PID: 14920"})
34
- elif str(actual_pid) == "0x3a48":
35
- score += 15
36
- details.append({"item": "PID 识别", "score": 15, "max_score": 30, "passed": False, "reason": "未将 PID 从十六进制转换为十进制"})
37
- else:
38
- details.append({"item": "PID 识别", "score": 0, "max_score": 30, "passed": False, "reason": f"PID 错误，识别为 {actual_pid}"})
39
 
40
- # 4. 检查 XID 是否正确 (30分)
41
- actual_xid = data.get("xid")
42
- if actual_xid == EXPECTED_XID:
43
- score += 30
44
- details.append({"item": "XID 识别", "score": 30, "max_score": 30, "passed": True, "reason": "成功识别事务 ID: 9948271"})
45
- else:
46
- details.append({"item": "XID 识别", "score": 0, "max_score": 30, "passed": False, "reason": f"XID 错误，识别为 {actual_xid}"})
47
-
48
- # 5. 检查是否存在多余字段或干扰项 (10分)
49
- # 要求只有 pid 和 xid
50
- allowed_keys = {"pid", "xid"}
51
- actual_keys = set(data.keys())
52
- if actual_keys == allowed_keys:
53
- score += 10
54
- details.append({"item": "字段精简度", "score": 10, "max_score": 10, "passed": True, "reason": "输出字段精准，无多余分析"})
55
- else:
56
- details.append({"item": "字段精简度", "score": 0, "max_score": 10, "passed": False, "reason": f"包含多余字段: {actual_keys - allowed_keys}"})
57
-
58
- except json.JSONDecodeError:
59
- details.append({"item": "JSON格式验证", "score": 0, "max_score": 20, "passed": False, "reason": "JSON 格式非法"})
60
- except Exception as e:
61
- details.append({"item": "异常错误", "score": 0, "max_score": 0, "passed": False, "reason": str(e)})
62
- else:
63
- details.append({"item": "文件检查", "score": 0, "max_score": 10, "passed": False, "reason": "ops/kill_target.json 不存在"})
64
-
65
- # 写入评分结果
66
- output_result = {
67
- "total_score": score,
68
- "details": details
69
  }
70
- with open("workplace_score.json", "w", encoding='utf-8') as f:
71
- json.dump(output_result, f, indent=2, ensure_ascii=False)
 
 
72
 
73
  if __name__ == "__main__":
74
- verify()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
9
+ workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'This multi-turn task is evaluated from the original per-turn verifier scripts stored in verifiers/*.jsonl. The task-local verify_workplace.py is intentionally a non-authoritative zero-score fallback.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'multi_turn',
24
+ "source_task_id": 'data_47',
25
+ "imported_task_id": 'data_persona_aligned_multi_turn_50_0011',
26
+ "action": 'task_local_turn_verifier_placeholder',
27
+ },
 
 
 
 
 
 
 
 
 
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
+ main()
persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0012/verify_workplace.py CHANGED
@@ -1,111 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import httpx
5
- from openai import OpenAI
6
-
7
- # Configuration for LLM Judge
8
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
9
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
10
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
11
 
12
- http_client = httpx.Client(verify=False)
13
- client = OpenAI(
14
- api_key=MOCK_API_KEY,
15
- base_url=MOCK_API_BASE,
16
- http_client=http_client
17
- )
18
 
19
- def llm_judge_content(prompt_text, file_content):
20
- """Unified interface for non-structured text validation."""
21
- try:
22
- response = client.chat.completions.create(
23
- model=MOCK_MODEL_NAME,
24
- messages=[
25
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
26
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
27
- ],
28
- temperature=0
29
- )
30
- return "yes" in response.choices[0].message.content.strip().lower()
31
- except Exception as e:
32
- print(f"LLM API Error: {e}")
33
- return False
34
-
35
- def verify():
36
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
37
- score = 0
38
- details = []
39
-
40
- report_path = os.path.join(workspace, "report/conflict_summary.json")
41
-
42
- # 1. Check file existence (10 points)
43
- if os.path.exists(report_path):
44
- score += 10
45
- details.append({"item": "Check report existence", "score": 10, "max_score": 10, "passed": True, "reason": "File exists."})
46
- else:
47
- details.append({"item": "Check report existence", "score": 0, "max_score": 10, "passed": False, "reason": "File not found."})
48
- # Cannot proceed without file
49
- final_output(score, details)
50
- return
51
-
52
- # 2. Check JSON validity (10 points)
53
- data = {}
54
- try:
55
- with open(report_path, 'r', encoding='utf-8') as f:
56
- data = json.load(f)
57
- score += 10
58
- details.append({"item": "JSON Format Validation", "score": 10, "max_score": 10, "passed": True, "reason": "Valid JSON format."})
59
- except Exception as e:
60
- details.append({"item": "JSON Format Validation", "score": 0, "max_score": 10, "passed": False, "reason": f"Invalid JSON: {e}"})
61
- final_output(score, details)
62
- return
63
-
64
- # 3. Check for the exactly three pieces of information constraint (10 points)
65
- # The prompt asks for "exactly three pieces of information"
66
- if len(data.keys()) == 3:
67
- score += 10
68
- details.append({"item": "Constraint: Exactly 3 fields", "score": 10, "max_score": 10, "passed": True, "reason": "Report contains exactly three keys."})
69
- else:
70
- details.append({"item": "Constraint: Exactly 3 fields", "score": 0, "max_score": 10, "passed": False, "reason": f"Expected 3 fields, found {len(data.keys())}."})
71
-
72
- # 4. Verify Library Name (25 points)
73
- # Use LLM to ensure flexible naming (fmt vs fmtlib)
74
- lib_name = str(next(iter(data.values()))) # Get first value as a placeholder if keys aren't named
75
- # Better to look for specific key logic, but prompt didn't define keys. Let's find the library name in the values.
76
- values_str = json.dumps(data)
77
- lib_correct = llm_judge_content("Does the following JSON content identify 'fmt' or 'fmtlib' as the conflicting library?", values_str)
78
- if lib_correct:
79
- score += 25
80
- details.append({"item": "Library Identification", "score": 25, "max_score": 25, "passed": True, "reason": "Identified 'fmtlib' correctly."})
81
- else:
82
- details.append({"item": "Library Identification", "score": 0, "max_score": 25, "passed": False, "reason": "Failed to identify 'fmtlib' as the conflicting library."})
83
-
84
- # 5. Verify Expected Version (20 points)
85
- # Expected version in manifest is 9.1.0
86
- if "9.1.0" in values_str:
87
- score += 20
88
- details.append({"item": "Expected Version Accuracy", "score": 20, "max_score": 20, "passed": True, "reason": "Found expected version 9.1.0."})
89
- else:
90
- details.append({"item": "Expected Version Accuracy", "score": 0, "max_score": 20, "passed": False, "reason": "Missing or incorrect expected version (should be 9.1.0)."})
91
-
92
- # 6. Verify Rogue Version (25 points)
93
- # Rogue version in log is 8.0.1
94
- if "8.0.1" in values_str:
95
- score += 25
96
- details.append({"item": "Rogue Version Accuracy", "score": 25, "max_score": 25, "passed": True, "reason": "Found rogue version 8.0.1."})
97
- else:
98
- details.append({"item": "Rogue Version Accuracy", "score": 0, "max_score": 25, "passed": False, "reason": "Missing or incorrect rogue version (should be 8.0.1)."})
99
-
100
- final_output(score, details)
101
-
102
- def final_output(score, details):
103
- output = {
104
- "total_score": min(100, score),
105
- "details": details
106
  }
107
- with open("workplace_score.json", "w", encoding="utf-8") as f:
108
- json.dump(output, f, indent=2, ensure_ascii=False)
 
 
109
 
110
  if __name__ == "__main__":
111
- verify()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'This multi-turn task is evaluated from the original per-turn verifier scripts stored in verifiers/*.jsonl. The task-local verify_workplace.py is intentionally a non-authoritative zero-score fallback.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'multi_turn',
24
+ "source_task_id": 'data_49',
25
+ "imported_task_id": 'data_persona_aligned_multi_turn_50_0012',
26
+ "action": 'task_local_turn_verifier_placeholder',
27
+ },
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
+ main()
persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0013/verify_workplace.py CHANGED
@@ -1,110 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
-
5
- def run_reference_logic(file_path):
6
- """
7
- 参考实现逻辑：解析原始 L2 数据并找出首个有效的买卖盘倒挂。
8
- """
9
- max_ts = -1
10
- result = None
11
-
12
- if not os.path.exists(file_path):
13
- return None
14
-
15
- with open(file_path, "r", encoding="utf-8") as f:
16
- for line in f:
17
- parts = line.strip().split('\x01')
18
- if len(parts) != 4:
19
- continue
20
-
21
- try:
22
- ts = int(parts[0])
23
- symbol = parts[1]
24
- bids_str = parts[2]
25
- asks_str = parts[3]
26
- except ValueError:
27
- continue
28
 
29
- # 严格单调递增检查
30
- if ts <= max_ts:
31
- continue
32
- max_ts = ts
33
 
34
- # 解析买盘最优价 (Bid[0])
35
- try:
36
- best_bid = float(bids_str.split('|')[0].split(':')[0])
37
- best_ask = float(asks_str.split('|')[0].split(':')[0])
38
- except (IndexError, ValueError):
39
- continue
40
-
41
- # 检查买卖盘倒挂 (Crossed Book)
42
- if best_bid >= best_ask:
43
- result = {"symbol": symbol, "timestamp": ts}
44
- break # 找到第一个符合条件的即可
45
- return result
46
-
47
- def main():
48
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
49
- target_json_path = os.path.join(workspace, "ops/target_replay.json")
50
- raw_data_path = os.path.join(workspace, "snapshots/l2_orderbook.dat")
51
-
52
- score = 0
53
- details = []
54
-
55
- # 1. 检查目标文件是否存在 (10分)
56
- if os.path.exists(target_json_path):
57
- score += 10
58
- details.append({"item": "检查结果文件是否存在", "score": 10, "max_score": 10, "passed": True, "reason": "文件 ops/target_replay.json 已生成"})
59
- else:
60
- details.append({"item": "检查结果文件是否存在", "score": 0, "max_score": 10, "passed": False, "reason": "文件 ops/target_replay.json 未找到"})
61
- # 写入最终结果并提前退出
62
- with open("workplace_score.json", "w") as f:
63
- json.dump({"total_score": 0, "details": details}, f)
64
- return
65
-
66
- # 2. 检查 JSON 格式及字段 (20分)
67
- agent_data = {}
68
- try:
69
- with open(target_json_path, "r") as f:
70
- agent_data = json.load(f)
71
- score += 10
72
- details.append({"item": "JSON 格式合法性", "score": 10, "max_score": 10, "passed": True, "reason": "JSON 解析成功"})
73
-
74
- # 检查是否包含核心字段（允许大小写差异，但必须逻辑清晰）
75
- keys = [k.lower() for k in agent_data.keys()]
76
- if any("symbol" in k for k in keys) and any("timestamp" in k for k in keys):
77
- score += 10
78
- details.append({"item": "核心字段检查", "score": 10, "max_score": 10, "passed": True, "reason": "包含 symbol 和 timestamp 相关字段"})
79
- else:
80
- details.append({"item": "核心字段检查", "score": 0, "max_score": 10, "passed": False, "reason": f"缺少必要字段，当前包含: {list(agent_data.keys())}"})
81
- except Exception as e:
82
- details.append({"item": "JSON 格式合法性", "score": 0, "max_score": 20, "passed": False, "reason": f"解析失败: {str(e)}"})
83
-
84
- # 3. 核心逻辑验证 (70分)
85
- gold_standard = run_reference_logic(raw_data_path)
86
- if not gold_standard:
87
- details.append({"item": "数据源验证", "score": 0, "max_score": 70, "passed": False, "reason": "无法从原始数据中计算出参考答案，请检查环境"})
88
- else:
89
- agent_symbol = None
90
- agent_ts = None
91
- # 模糊匹配 Key
92
- for k, v in agent_data.items():
93
- if "symbol" in k.lower(): agent_symbol = v
94
- if "timestamp" in k.lower() or "ts" in k.lower(): agent_ts = v
95
-
96
- if str(agent_symbol) == str(gold_standard["symbol"]) and str(agent_ts) == str(gold_standard["timestamp"]):
97
- score += 70
98
- details.append({"item": "计算结果准确性", "score": 70, "max_score": 70, "passed": True, "reason": "成功定位到唯一的有效倒挂记录：FAT_FINGER_X"})
99
- elif str(agent_symbol) == "TRAP_SYM":
100
- score += 20
101
- details.append({"item": "计算结果准确性", "score": 20, "max_score": 70, "passed": False, "reason": "错误！Agent 抓取了被时间戳倒挂过滤掉的陷阱数据 (TRAP_SYM)"})
102
- else:
103
- details.append({"item": "计算结果准确性", "score": 0, "max_score": 70, "passed": False, "reason": f"结果不匹配。期望: {gold_standard}, 实际: {agent_data}"})
104
 
105
- # 写入最终总分
106
- with open("workplace_score.json", "w") as f:
107
- json.dump({"total_score": score, "details": details}, f, indent=2)
108
 
109
  if __name__ == "__main__":
110
  main()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6
 
 
 
 
 
7
 
8
+ def main() -> None:
 
 
 
 
 
 
 
 
 
 
 
 
 
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'This multi-turn task is evaluated from the original per-turn verifier scripts stored in verifiers/*.jsonl. The task-local verify_workplace.py is intentionally a non-authoritative zero-score fallback.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'multi_turn',
24
+ "source_task_id": 'data_05',
25
+ "imported_task_id": 'data_persona_aligned_multi_turn_50_0013',
26
+ "action": 'task_local_turn_verifier_placeholder',
27
+ },
28
+ }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
32
 
 
 
 
33
 
34
  if __name__ == "__main__":
35
  main()
persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0014/verify_workplace.py CHANGED
@@ -1,93 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
-
5
- def verify():
6
- workspace = sys.argv[1] if len(sys.argv) > 1 else "."
7
- report_path = os.path.join(workspace, "report/root_cause.json")
8
- score = 0
9
- details = []
10
-
11
- # 1. Check Directory and File Existence (10 points)
12
- if os.path.exists(os.path.join(workspace, "report")):
13
- score += 5
14
- details.append({"item": "检查报告目录", "score": 5, "max_score": 5, "passed": True, "reason": "目录 report 存在"})
15
- else:
16
- details.append({"item": "检查报告目录", "score": 0, "max_score": 5, "passed": False, "reason": "目录 report 不存在"})
17
-
18
- if os.path.exists(report_path):
19
- score += 5
20
- details.append({"item": "检查报告文件是否存在", "score": 5, "max_score": 5, "passed": True, "reason": "文件 root_cause.json 存在"})
21
- else:
22
- details.append({"item": "检查报告文件是否存在", "score": 0, "max_score": 5, "passed": False, "reason": "文件 root_cause.json 不存在"})
23
- # If the file doesn't exist, we can't perform further checks
24
- save_results(score, details)
25
- return
26
-
27
- # 2. JSON Validity and Structure (20 points)
28
- try:
29
- with open(report_path, 'r', encoding='utf-8') as f:
30
- data = json.load(f)
31
- score += 10
32
- details.append({"item": "检查JSON格式合法性", "score": 10, "max_score": 10, "passed": True, "reason": "JSON解析成功"})
33
-
34
- required_keys = ["device_address", "register_address", "illegal_value"]
35
- missing_keys = [k for k in required_keys if k not in data]
36
- if not missing_keys:
37
- score += 10
38
- details.append({"item": "检查JSON关键字段", "score": 10, "max_score": 10, "passed": True, "reason": "包含所有必需字段"})
39
- else:
40
- details.append({"item": "检查JSON关键字段", "score": 0, "max_score": 10, "passed": False, "reason": f"缺失字段: {missing_keys}"})
41
- except Exception as e:
42
- details.append({"item": "检查JSON格式合法性", "score": 0, "max_score": 20, "passed": False, "reason": f"JSON解析失败: {str(e)}"})
43
- save_results(score, details)
44
- return
45
 
46
- # 3. Content Accuracy (70 points)
47
- # Target Values based on env_builder.py:
48
- # device_address: 0x5C, register_address: 0x10, illegal_value: 0x4B
49
-
50
- def normalize_hex(val):
51
- if not isinstance(val, str): return None
52
- try:
53
- return hex(int(val, 16)).lower()
54
- except:
55
- return None
56
 
57
- dev_addr = normalize_hex(data.get("device_address"))
58
- reg_addr = normalize_hex(data.get("register_address"))
59
- ill_val = normalize_hex(data.get("illegal_value"))
60
-
61
- # Device Address (20 points)
62
- if dev_addr == "0x5c":
63
- score += 20
64
- details.append({"item": "验证设备地址 (device_address)", "score": 20, "max_score": 20, "passed": True, "reason": "正确识别 PMIC 地址 0x5C"})
65
- else:
66
- details.append({"item": "验证设备地址 (device_address)", "score": 0, "max_score": 20, "passed": False, "reason": f"预期 0x5C, 实际得到 {data.get('device_address')}"})
67
-
68
- # Register Address (20 points)
69
- if reg_addr == "0x10":
70
- score += 20
71
- details.append({"item": "验证寄存器地址 (register_address)", "score": 20, "max_score": 20, "passed": True, "reason": "正确识别核心电压寄存器 0x10"})
72
- else:
73
- details.append({"item": "验证寄存器地址 (register_address)", "score": 0, "max_score": 20, "passed": False, "reason": f"预期 0x10, 实际得到 {data.get('register_address')}"})
74
-
75
- # Illegal Value (30 points)
76
- if ill_val == "0x4b":
77
- score += 30
78
- details.append({"item": "验证非法写入值 (illegal_value)", "score": 30, "max_score": 30, "passed": True, "reason": "正确锁定导致崩溃的非法值 0x4B (超过 0x3F)"})
79
- else:
80
- details.append({"item": "验证非法写入值 (illegal_value)", "score": 0, "max_score": 30, "passed": False, "reason": f"预期 0x4B, 实际得到 {data.get('illegal_value')}"})
81
-
82
- save_results(score, details)
83
-
84
- def save_results(score, details):
85
- output = {
86
- "total_score": score,
87
- "details": details
88
  }
89
- with open("workplace_score.json", "w", encoding="utf-8") as f:
90
- json.dump(output, f, indent=2, ensure_ascii=False)
 
 
91
 
92
  if __name__ == "__main__":
93
- verify()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
9
+ workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'This multi-turn task is evaluated from the original per-turn verifier scripts stored in verifiers/*.jsonl. The task-local verify_workplace.py is intentionally a non-authoritative zero-score fallback.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'multi_turn',
24
+ "source_task_id": 'data_87',
25
+ "imported_task_id": 'data_persona_aligned_multi_turn_50_0014',
26
+ "action": 'task_local_turn_verifier_placeholder',
27
+ },
 
 
 
 
 
 
 
 
 
 
 
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
+ main()
persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0015/verify_workplace.py CHANGED
@@ -1,114 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import httpx
5
- from openai import OpenAI
6
-
7
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
8
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
9
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
10
-
11
- # 初始化客户端，必须关闭 SSL 验证
12
- http_client = httpx.Client(verify=False)
13
- client = OpenAI(
14
- api_key=MOCK_API_KEY,
15
- base_url=MOCK_API_BASE,
16
- http_client=http_client
17
- )
18
 
19
- def llm_judge_content(prompt_text, file_content):
20
- # 此函数为检测非结构化文本的统一接口
21
- try:
22
- response = client.chat.completions.create(
23
- model=MOCK_MODEL_NAME,
24
- messages=[
25
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
26
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
27
- ],
28
- temperature=0
29
- )
30
- return "yes" in response.choices[0].message.content.strip().lower()
31
- except Exception as e:
32
- print(f"LLM API Error: {e}")
33
- return False
34
 
35
- def verify():
36
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
37
- target_file = os.path.join(workspace, "optimizations", "target_gates.json")
38
-
39
- details = []
40
- total_score = 0
41
-
42
- # 1. 检查目标文件是否存在 (10分)
43
- if os.path.exists(target_file):
44
- details.append({"item": "检查目标文件 target_gates.json 是否存在", "score": 10, "max_score": 10, "passed": True, "reason": "文件已建立"})
45
- total_score += 10
46
- else:
47
- details.append({"item": "检查目标文件 target_gates.json 是否存在", "score": 0, "max_score": 10, "passed": False, "reason": "文件不存在，未按要求输出"})
48
- write_score(total_score, details)
49
- return
50
-
51
- # 2. 检查 JSON 格式合法性 (10分)
52
- # 此处严禁对结构化数据进行模糊匹配，必须原生解析
53
- try:
54
- with open(target_file, "r", encoding="utf-8") as f:
55
- data = json.load(f)
56
- details.append({"item": "检查 JSON 格式合法性", "score": 10, "max_score": 10, "passed": True, "reason": "可成功解析为 JSON"})
57
- total_score += 10
58
- except Exception as e:
59
- details.append({"item": "检查 JSON 格式合法性", "score": 0, "max_score": 10, "passed": False, "reason": f"非合法 JSON 格式，解析报错: {e}"})
60
- write_score(total_score, details)
61
- return
62
-
63
- # 3. 检查 Schema 数据结构合规性 (10分)
64
- # 题目明确要求输出 3 个逻辑门的 ID
65
- if isinstance(data, list) and len(data) == 3 and all(isinstance(x, str) for x in data):
66
- details.append({"item": "检查 Schema（含有3个字符串的数组）", "score": 10, "max_score": 10, "passed": True, "reason": "结构符合要求：一个包含3个字符串元素的列表"})
67
- total_score += 10
68
- else:
69
- details.append({"item": "检查 Schema（含有3个字符串的数组）", "score": 0, "max_score": 10, "passed": False, "reason": f"未返回仅包含 3 个字符串的数组。当前数据：{data}"})
70
- # 结构不对扣光后续分数，结束验证
71
- write_score(total_score, details)
72
- return
73
-
74
- # 4. 检查内容：命中率 (30分，每个关键 ID 10分)
75
- # 根据底层注入规则：数据量最大的 3 个门固定为 F9A1, F9A2, F9A3
76
- expected_gates = {"GATE_F9A1", "GATE_F9A2", "GATE_F9A3"}
77
- matched = set(data).intersection(expected_gates)
78
- hit_count = len(matched)
79
- hit_score = hit_count * 10
80
-
81
- if hit_count == 3:
82
- details.append({"item": "检查提取目标门 ID 的精确度", "score": 30, "max_score": 30, "passed": True, "reason": "完美找出所有 3 个异常通信量的逻辑门"})
83
- else:
84
- missing = expected_gates - set(data)
85
- details.append({"item": "检查提取目标门 ID 的精确度", "score": hit_score, "max_score": 30, "passed": False, "reason": f"找到了 {hit_count} 个异常逻辑门, 缺失 {missing}"})
86
- total_score += hit_score
87
-
88
- # 5. 检查内容：排序正确性 (40分)
89
- # 数据量排序：GATE_F9A1(3500B) > GATE_F9A2(2800B) > GATE_F9A3(2100B)
90
- expected_order = ["GATE_F9A1", "GATE_F9A2", "GATE_F9A3"]
91
- if data == expected_order:
92
- details.append({"item": "检查数组降序排序正确性", "score": 40, "max_score": 40, "passed": True, "reason": "元素严格按数据载荷大小从大到小排列"})
93
- total_score += 40
94
- else:
95
- # 如果找齐了三个，但排序错误，可给出部分分数 10 分以作梯度区分，如果没找齐则得 0 分。
96
- if hit_count == 3:
97
- details.append({"item": "检查数组降序排序正确性", "score": 10, "max_score": 40, "passed": False, "reason": f"找出了所有门但排序错误，期望 {expected_order}，实际 {data}"})
98
- total_score += 10
99
- else:
100
- details.append({"item": "检查数组降序排序正确性", "score": 0, "max_score": 40, "passed": False, "reason": "由于存在找错的门，排序亦判定为完全失败"})
101
-
102
- write_score(total_score, details)
103
-
104
- def write_score(total_score, details):
105
- res = {
106
- "total_score": total_score,
107
- "details": details
108
  }
109
- # 确保写入工作目录下的 workplace_score.json
110
- with open("workplace_score.json", "w", encoding="utf-8") as f:
111
- json.dump(res, f, ensure_ascii=False, indent=2)
 
112
 
113
  if __name__ == "__main__":
114
- verify()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'This multi-turn task is evaluated from the original per-turn verifier scripts stored in verifiers/*.jsonl. The task-local verify_workplace.py is intentionally a non-authoritative zero-score fallback.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'multi_turn',
24
+ "source_task_id": 'data_73',
25
+ "imported_task_id": 'data_persona_aligned_multi_turn_50_0015',
26
+ "action": 'task_local_turn_verifier_placeholder',
27
+ },
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
+ main()
persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0016/verify_workplace.py CHANGED
@@ -1,160 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import re
5
- import httpx
6
- from openai import OpenAI
7
 
8
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
9
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
10
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
11
 
12
- # 初始化客户端，必须关闭 SSL 验证
13
- http_client = httpx.Client(verify=False)
14
- client = OpenAI(
15
- api_key=MOCK_API_KEY,
16
- base_url=MOCK_API_BASE,
17
- http_client=http_client
18
- )
19
-
20
- def llm_judge_content(prompt_text, file_content):
21
- try:
22
- response = client.chat.completions.create(
23
- model=MOCK_MODEL_NAME,
24
- messages=[
25
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
26
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
27
- ],
28
- temperature=0
29
- )
30
- return "yes" in response.choices[0].message.content.strip().lower()
31
- except Exception as e:
32
- print(f"LLM API Error: {e}")
33
- return False
34
-
35
- def verify():
36
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
37
- target_file = os.path.join(workspace, "processed", "clean_traj_ids.txt")
38
-
39
- score_details = []
40
- total_score = 0
41
-
42
- # 1. 检查目标文件是否存在 (15分)
43
- file_exists = os.path.exists(target_file) and os.path.isfile(target_file)
44
- if file_exists:
45
- score_details.append({"item": "检查目标文件是否存在", "score": 15, "max_score": 15, "passed": True, "reason": "成功找到 `processed/clean_traj_ids.txt`"})
46
- total_score += 15
47
- else:
48
- score_details.append({"item": "检查目标文件是否存在", "score": 0, "max_score": 15, "passed": False, "reason": "未找到目标文件 `processed/clean_traj_ids.txt`"})
49
-
50
- extracted_ids = []
51
- pure_format = False
52
-
53
- if file_exists:
54
- try:
55
- with open(target_file, "r", encoding="utf-8") as f:
56
- lines = f.readlines()
57
-
58
- # 清理空行并解析
59
- raw_lines = [line.strip() for line in lines if line.strip()]
60
-
61
- # 2. 检查输出纯净度 (10分)
62
- # 要求：别整那些没用的报告，只要 ID。所有的非空行应当全都是形如 T-xxxx 的格式。
63
- pure_format = all(re.match(r'^T-\d+$', line) for line in raw_lines)
64
- if pure_format and len(raw_lines) > 0:
65
- score_details.append({"item": "检查输出纯净度", "score": 10, "max_score": 10, "passed": True, "reason": "输出无冗余报告废话，每行格式均为合法的 ID"})
66
- total_score += 10
67
- else:
68
- score_details.append({"item": "检查输出纯净度", "score": 0, "max_score": 10, "passed": False, "reason": "输出中混杂了自然语言报告、无效格式或文件全空"})
69
-
70
- # 尝试提取出里面看起来像ID的串用于进一步逻辑验证
71
- extracted_ids = []
72
- for line in raw_lines:
73
- # 若包含自然语言，则利用正则从中强行提取 T-xxx 评估实质清洗逻辑
74
- matches = re.findall(r'T-\d+', line)
75
- extracted_ids.extend(matches)
76
-
77
- except Exception as e:
78
- score_details.append({"item": "检查输出纯净度", "score": 0, "max_score": 10, "passed": False, "reason": f"文件读取异常: {e}"})
79
- else:
80
- score_details.append({"item": "检查输出纯净度", "score": 0, "max_score": 10, "passed": False, "reason": "文件不存在，跳过纯净度检查"})
81
-
82
- extracted_set = set(extracted_ids)
83
-
84
- # 3. 检查乱码与损坏数据过滤能力 (15分)
85
- # T-1005 (json未闭合) 和 T-1006 (十六进制乱码前缀) 不应存在
86
- if file_exists:
87
- if "T-1005" not in extracted_set and "T-1006" not in extracted_set:
88
- score_details.append({"item": "过滤损坏的JSON与乱码数据", "score": 15, "max_score": 15, "passed": True, "reason": "正确剔除了无法被正常解析的 T-1005 和 T-1006"})
89
- total_score += 15
90
- else:
91
- score_details.append({"item": "过滤损坏的JSON与乱码数据", "score": 0, "max_score": 15, "passed": False, "reason": "未正确拦截格式残缺或含乱码前缀的脏数据"})
92
- else:
93
- score_details.append({"item": "过滤损坏的JSON与乱码数据", "score": 0, "max_score": 15, "passed": False, "reason": "文件不存在，该项记 0 分"})
94
-
95
- # 4. 检查 Token 截断数据的过滤能力 (15分)
96
- # T-1003 (finish_reason = length) 不应存在
97
- if file_exists:
98
- if "T-1003" not in extracted_set:
99
- score_details.append({"item": "过滤严重截断数据", "score": 15, "max_score": 15, "passed": True, "reason": "正确剔除了因为 max_tokens 截断（length）的残缺数据 T-1003"})
100
- total_score += 15
101
- else:
102
- score_details.append({"item": "过滤严重截断数据", "score": 0, "max_score": 15, "passed": False, "reason": "未正确拦截 finish_reason 为 length 的数据"})
103
- else:
104
- score_details.append({"item": "过滤严重截断数据", "score": 0, "max_score": 15, "passed": False, "reason": "文件不存在，该项记 0 分"})
105
-
106
- # 5. 检查 Agent 死循环数据的过滤能力 (15分)
107
- # T-1002 (连续相同的 tool calls) 不应存在
108
- if file_exists:
109
- if "T-1002" not in extracted_set:
110
- score_details.append({"item": "过滤陷入死循环的工具调用数据", "score": 15, "max_score": 15, "passed": True, "reason": "正确剔除了陷入无意义重复调用的 T-1002 数据"})
111
- total_score += 15
112
- else:
113
- score_details.append({"item": "过滤陷入死循环的工具调用数据", "score": 0, "max_score": 15, "passed": False, "reason": "未正确拦截呈现死循环特征的轨迹数据 T-1002"})
114
- else:
115
- score_details.append({"item": "过滤陷入死循环的工具调用数据", "score": 0, "max_score": 15, "passed": False, "reason": "文件不存在，该项记 0 分"})
116
-
117
- # 6. 检查健康轨迹的全量留存与拒答幻觉情况 (30分)
118
- # 正确的数据集应当只包含：T-1001, T-1004, T-1007
119
- if file_exists:
120
- expected_ids = {"T-1001", "T-1004", "T-1007"}
121
- missing_ids = expected_ids - extracted_set
122
- # 计算提取列表中多出的且不在剔除名单里的虚构 ID (幻觉)
123
- dirty_ids = {"T-1002", "T-1003", "T-1005", "T-1006"}
124
- hallucinated_ids = extracted_set - expected_ids - dirty_ids
125
-
126
- health_score = 30
127
- penalties = []
128
-
129
- if len(missing_ids) > 0:
130
- penalty = len(missing_ids) * 10
131
- health_score -= penalty
132
- penalties.append(f"遗漏了合法数据 {missing_ids}，扣 {penalty} 分")
133
-
134
- if len(hallucinated_ids) > 0:
135
- penalty = len(hallucinated_ids) * 10
136
- health_score -= penalty
137
- penalties.append(f"捏造了不存在的数据 ID {hallucinated_ids}，扣 {penalty} 分")
138
-
139
- health_score = max(0, health_score)
140
-
141
- if health_score == 30:
142
- score_details.append({"item": "精准留存合法数据与幻觉抑制", "score": 30, "max_score": 30, "passed": True, "reason": "完美找出了所有的健康数据 T-1001, T-1004, T-1007，且无捏造或多余数据！"})
143
- else:
144
- score_details.append({"item": "精准留存合法数据与幻觉抑制", "score": health_score, "max_score": 30, "passed": False, "reason": "; ".join(penalties)})
145
-
146
- total_score += health_score
147
- else:
148
- score_details.append({"item": "精准留存合法数据与幻觉抑制", "score": 0, "max_score": 30, "passed": False, "reason": "文件不存在，该项记 0 分"})
149
-
150
- # 保存评分文件
151
- score_data = {
152
- "total_score": total_score,
153
- "details": score_details
154
  }
155
-
156
- with open(os.path.join(workspace, "workplace_score.json"), "w", encoding="utf-8") as f:
157
- json.dump(score_data, f, ensure_ascii=False, indent=2)
 
158
 
159
  if __name__ == "__main__":
160
- verify()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
6
 
 
 
 
7
 
8
+ def main() -> None:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'This multi-turn task is evaluated from the original per-turn verifier scripts stored in verifiers/*.jsonl. The task-local verify_workplace.py is intentionally a non-authoritative zero-score fallback.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'multi_turn',
24
+ "source_task_id": 'data_07',
25
+ "imported_task_id": 'data_persona_aligned_multi_turn_50_0016',
26
+ "action": 'task_local_turn_verifier_placeholder',
27
+ },
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
+ main()
persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0017/verify_workplace.py CHANGED
@@ -1,124 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import httpx
5
- from openai import OpenAI
6
-
7
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
8
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
9
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
10
 
11
- # 初始化客户端，必须关闭 SSL 验证
12
- http_client = httpx.Client(verify=False)
13
- client = OpenAI(
14
- api_key=MOCK_API_KEY,
15
- base_url=MOCK_API_BASE,
16
- http_client=http_client
17
- )
18
 
19
- def llm_judge_content(prompt_text, file_content):
20
- # 此函数为检测非结构化文本的统一接口
21
- try:
22
- response = client.chat.completions.create(
23
- model=MOCK_MODEL_NAME,
24
- messages=[
25
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
26
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
27
- ],
28
- temperature=0
29
- )
30
- return "yes" in response.choices[0].message.content.strip().lower()
31
- except Exception as e:
32
- print(f"LLM API Error: {e}")
33
- return False
34
-
35
- def main():
36
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
37
- report_file = os.path.join(workspace, "report", "hacker.json")
38
-
39
- score = 0
40
- details = []
41
-
42
- # 1. 检查目标目录存在性 (5分)
43
- has_dir = os.path.isdir(os.path.join(workspace, "report"))
44
- if has_dir:
45
- score += 5
46
- details.append({"item": "检查 report 目录是否存在", "score": 5, "max_score": 5, "passed": True, "reason": "report 目录存在"})
47
- else:
48
- details.append({"item": "检查 report 目录是否存在", "score": 0, "max_score": 5, "passed": False, "reason": "report 目录不存在"})
49
-
50
- # 2. 检查结果文件存在性 (15分)
51
- has_file = os.path.isfile(report_file)
52
- if has_file:
53
- score += 15
54
- details.append({"item": "检查 hacker.json 文件是否存在", "score": 15, "max_score": 15, "passed": True, "reason": "hacker.json 文件存在"})
55
- else:
56
- details.append({"item": "检查 hacker.json 文件是否存在", "score": 0, "max_score": 15, "passed": False, "reason": "hacker.json 文件不存在"})
57
-
58
- if not has_file:
59
- save_score(score, details)
60
- return
61
-
62
- # 3. 检查文件格式合法性 (10分)
63
- try:
64
- with open(report_file, "r", encoding="utf-8") as f:
65
- data = json.load(f)
66
- score += 10
67
- details.append({"item": "文件 JSON 格式合法性", "score": 10, "max_score": 10, "passed": True, "reason": "成功解析为 JSON 格式"})
68
- except Exception as e:
69
- details.append({"item": "文件 JSON 格式合法性", "score": 0, "max_score": 10, "passed": False, "reason": f"无法解析为 JSON，解析失败: {str(e)}"})
70
- save_score(score, details)
71
- return
72
-
73
- # 4. 字段规范与幻觉严查 (10分)
74
- if not isinstance(data, dict):
75
- details.append({"item": "字段规范检查", "score": 0, "max_score": 10, "passed": False, "reason": "JSON 根节点不是对象(dict)结构"})
76
- else:
77
- keys = list(data.keys())
78
- expected_keys = {"hacker_address", "exploit_tx_hash"}
79
- if set(keys) == expected_keys:
80
- score += 10
81
- details.append({"item": "字段规范检查", "score": 10, "max_score": 10, "passed": True, "reason": "字段完全匹配要求，无多余捏造字段"})
82
- elif expected_keys.issubset(set(keys)):
83
- score += 5
84
- details.append({"item": "字段规范检查", "score": 5, "max_score": 10, "passed": False, "reason": "包含目标字段，但存在捏造的多余字段，部分扣分"})
85
- else:
86
- missing = expected_keys - set(keys)
87
- details.append({"item": "字段规范检查", "score": 0, "max_score": 10, "passed": False, "reason": f"缺失必要字段: {missing}"})
88
-
89
- # 5. 黑客原始地址提取准确性 (30分)
90
- ans_addr = "0xbadc0ffeebadc0ffeebadc0ffeebadc0ffeebadc"
91
- if isinstance(data, dict) and "hacker_address" in data:
92
- addr = str(data["hacker_address"]).strip().lower()
93
- if addr == ans_addr:
94
- score += 30
95
- details.append({"item": "黑客地址正确性", "score": 30, "max_score": 30, "passed": True, "reason": "精准定位并提取了对应的 hacker_address"})
96
- else:
97
- details.append({"item": "黑客地址正确性", "score": 0, "max_score": 30, "passed": False, "reason": f"hacker_address 错误。期望: {ans_addr}，实际: {addr}"})
98
- else:
99
- details.append({"item": "黑客地址正确性", "score": 0, "max_score": 30, "passed": False, "reason": "无法验证，因文件内缺失 hacker_address 字段"})
100
-
101
- # 6. 致命交易哈希提取准确性 (30分)
102
- ans_hash = "0xdeadbeef999999999999999999999999999999999999999999999999deadbeef"
103
- if isinstance(data, dict) and "exploit_tx_hash" in data:
104
- tx_hash = str(data["exploit_tx_hash"]).strip().lower()
105
- if tx_hash == ans_hash:
106
- score += 30
107
- details.append({"item": "攻击交易哈希正确性", "score": 30, "max_score": 30, "passed": True, "reason": "精准提取了对应的 exploit_tx_hash"})
108
- else:
109
- details.append({"item": "攻击交易哈希正确性", "score": 0, "max_score": 30, "passed": False, "reason": f"exploit_tx_hash 错误。期望: {ans_hash}，实际: {tx_hash}"})
110
- else:
111
- details.append({"item": "攻击交易哈希正确性", "score": 0, "max_score": 30, "passed": False, "reason": "无法验证，因文件内缺失 exploit_tx_hash 字段"})
112
-
113
- save_score(score, details)
114
-
115
- def save_score(score, details):
116
- res = {
117
- "total_score": score,
118
- "details": details
119
  }
120
- with open("workplace_score.json", "w", encoding="utf-8") as f:
121
- json.dump(res, f, ensure_ascii=False, indent=2)
 
 
122
 
123
  if __name__ == "__main__":
124
  main()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'This multi-turn task is evaluated from the original per-turn verifier scripts stored in verifiers/*.jsonl. The task-local verify_workplace.py is intentionally a non-authoritative zero-score fallback.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'multi_turn',
24
+ "source_task_id": 'data_83',
25
+ "imported_task_id": 'data_persona_aligned_multi_turn_50_0017',
26
+ "action": 'task_local_turn_verifier_placeholder',
27
+ },
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
  main()
persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0018/verify_workplace.py CHANGED
@@ -1,142 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import re
5
- import httpx
6
- from openai import OpenAI
7
-
8
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
9
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
10
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
11
 
12
- # 初始化客户端，必须关闭 SSL 验证
13
- http_client = httpx.Client(verify=False)
14
- client = OpenAI(
15
- api_key=MOCK_API_KEY,
16
- base_url=MOCK_API_BASE,
17
- http_client=http_client
18
- )
19
 
20
- def llm_judge_content(prompt_text, file_content):
21
- # 此函数为检测非结构化文本的统一接口
22
- try:
23
- response = client.chat.completions.create(
24
- model=MOCK_MODEL_NAME,
25
- messages=[
26
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
27
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
28
- ],
29
- temperature=0
30
- )
31
- return "yes" in response.choices[0].message.content.strip().lower()
32
- except Exception as e:
33
- print(f"LLM API Error: {e}")
34
- return False
35
-
36
- def get_ground_truth(workspace):
37
- """
38
- 沙盒内探针自己执行硬核逻辑，重新解析数据计算 Ground Truth，避免假阴性/假阳性。
39
- """
40
- can_log = os.path.join(workspace, "chassis_can.log")
41
- radar_json = os.path.join(workspace, "sensor_data", "radar_track.json")
42
-
43
- if not os.path.exists(can_log) or not os.path.exists(radar_json):
44
- return set()
45
-
46
- aeb_timestamps = []
47
- with open(can_log, "r", encoding="utf-8") as f:
48
- for line in f:
49
- # 必须满足双重条件: 刹车CAN ID = 0x2B0, PAYLOAD 前两字节 = FF 01
50
- if "MSG_ID:0x2B0" in line and "PAYLOAD:[FF 01" in line:
51
- m = re.search(r"<(\d+)>", line)
52
- if m:
53
- aeb_timestamps.append(int(m.group(1)))
54
-
55
- truth_ids = set()
56
- with open(radar_json, "r", encoding="utf-8") as f:
57
- radar_data = json.load(f)
58
-
59
- frames = radar_data.get("data_stream", {}).get("radar_front_center", {}).get("frames", [])
60
- for frame in frames:
61
- stamp_ms = frame.get("header", {}).get("stamp_ms", 0)
62
- # 严密的时间戳对齐：雷达比底盘快 1500ms
63
- if (stamp_ms - 1500) in aeb_timestamps:
64
- objects = frame.get("payload", {}).get("tracked_entities", {}).get("radar_objects", [])
65
- for obj in objects:
66
- rcs = obj.get("attributes", {}).get("rcs_dbsm", 999.0)
67
- conf = obj.get("attributes", {}).get("track_confidence", 999)
68
- # 必须满足 rcs < 5.0 且 confidence < 60
69
- if rcs < 5.0 and conf < 60:
70
- tid = obj.get("metadata", {}).get("track_id", "")
71
- if tid:
72
- truth_ids.add(tid)
73
-
74
- return truth_ids
75
-
76
- def main():
77
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
78
- target_file = os.path.join(workspace, "analysis", "ghost_ids.json")
79
-
80
- details = []
81
- total_score = 0
82
-
83
- # 1. 验证目标文件存在性 (10分)
84
- if os.path.exists(target_file):
85
- total_score += 10
86
- details.append({"item": "检查目标文件是否存在", "score": 10, "max_score": 10, "passed": True, "reason": "文件 analysis/ghost_ids.json 存在"})
87
- else:
88
- details.append({"item": "检查目标文件是否存在", "score": 0, "max_score": 10, "passed": False, "reason": "文件 analysis/ghost_ids.json 不存在"})
89
-
90
- # 2. 验证结构纯净性 (20分)
91
- # 绝对禁止使用正则去匹配结构化结果，必须使用 json 库严格解析
92
- agent_ids = []
93
- is_valid_format = False
94
- if os.path.exists(target_file):
95
- try:
96
- with open(target_file, "r", encoding="utf-8") as f:
97
- data = json.load(f)
98
- if isinstance(data, list) and all(isinstance(i, str) for i in data):
99
- is_valid_format = True
100
- agent_ids = data
101
- total_score += 20
102
- details.append({"item": "JSON格式规范性验证", "score": 20, "max_score": 20, "passed": True, "reason": "是一个纯净的字符串数组"})
103
- else:
104
- details.append({"item": "JSON格式规范性验证", "score": 0, "max_score": 20, "passed": False, "reason": "结构错误，不是纯净的字符串数组"})
105
- except json.JSONDecodeError:
106
- details.append({"item": "JSON格式规范性验证", "score": 0, "max_score": 20, "passed": False, "reason": "非法的JSON文件"})
107
- else:
108
- details.append({"item": "JSON格式规范性验证", "score": 0, "max_score": 20, "passed": False, "reason": "文件缺失，无法验证"})
109
-
110
- # 3. 数据精准度 (70分)
111
- if is_valid_format:
112
- truth_ids = get_ground_truth(workspace)
113
- agent_set = set(agent_ids)
114
-
115
- if not truth_ids:
116
- # 如果极端情况环境加载异常，这里进行容错
117
- details.append({"item": "验证提取的 ID 准确性", "score": 0, "max_score": 70, "passed": False, "reason": "Ground Truth 数据生成错误，请检查环境"})
118
- else:
119
- intersection = agent_set.intersection(truth_ids)
120
- false_positives = agent_set - truth_ids
121
- false_negatives = truth_ids - agent_set
122
-
123
- union_len = len(agent_set.union(truth_ids))
124
- # 使用严格的 Jaccard 相似度来反映 F1 维度的惩罚机制，有捏造、漏报均会急剧降分
125
- data_score = int(70 * (len(intersection) / union_len)) if union_len > 0 else 0
126
-
127
- total_score += data_score
128
- passed = (data_score == 70)
129
- reason = f"精准度检查完成。正确提取: {len(intersection)}项, 漏报: {len(false_negatives)}项, 误报(幻觉/条件错误): {len(false_positives)}项"
130
- details.append({"item": "验证提取的 ID 准确性", "score": data_score, "max_score": 70, "passed": passed, "reason": reason})
131
- else:
132
- details.append({"item": "验证提取的 ID 准确性", "score": 0, "max_score": 70, "passed": False, "reason": "由于文件不存在或格式不合规，跳过数据校验"})
133
 
134
- # 统分写入
135
- with open(os.path.join(workspace, "workplace_score.json"), "w", encoding="utf-8") as f:
136
- json.dump({
137
- "total_score": total_score,
138
- "details": details
139
- }, f, indent=2, ensure_ascii=False)
140
 
141
  if __name__ == "__main__":
142
  main()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'This multi-turn task is evaluated from the original per-turn verifier scripts stored in verifiers/*.jsonl. The task-local verify_workplace.py is intentionally a non-authoritative zero-score fallback.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'multi_turn',
24
+ "source_task_id": 'data_08',
25
+ "imported_task_id": 'data_persona_aligned_multi_turn_50_0018',
26
+ "action": 'task_local_turn_verifier_placeholder',
27
+ },
28
+ }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
32
 
 
 
 
 
 
 
33
 
34
  if __name__ == "__main__":
35
  main()
persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0019/verify_workplace.py CHANGED
@@ -1,107 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import httpx
5
- from openai import OpenAI
6
-
7
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
8
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
9
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
10
-
11
- # 初始化客户端，必须关闭 SSL 验证
12
- http_client = httpx.Client(verify=False)
13
- client = OpenAI(
14
- api_key=MOCK_API_KEY,
15
- base_url=MOCK_API_BASE,
16
- http_client=http_client
17
- )
18
 
19
- def llm_judge_content(prompt_text, file_content):
20
- # 此函数为检测非结构化文本的统一接口
21
- try:
22
- response = client.chat.completions.create(
23
- model=MOCK_MODEL_NAME,
24
- messages=[
25
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
26
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
27
- ],
28
- temperature=0
29
- )
30
- return "yes" in response.choices[0].message.content.strip().lower()
31
- except Exception as e:
32
- print(f"LLM API Error: {e}")
33
- return False
34
 
35
- def verify_target(workspace):
36
- score = 0
37
- details = []
38
-
39
- # 1. 检查目标目录是否存在 (10 分)
40
- fix_list_dir = os.path.join(workspace, "fix_list")
41
- if os.path.isdir(fix_list_dir):
42
- score += 10
43
- details.append({"item": "检查 fix_list 目录是否存在", "score": 10, "max_score": 10, "passed": True, "reason": "目录 fix_list 成功创建"})
44
- else:
45
- details.append({"item": "检查 fix_list 目录是否存在", "score": 0, "max_score": 10, "passed": False, "reason": "目录 fix_list 不存在"})
46
-
47
- # 2. 检查结果文件是否存在 (20 分)
48
- target_file = os.path.join(workspace, "fix_list", "target.json")
49
- if os.path.isfile(target_file):
50
- score += 20
51
- details.append({"item": "检查 target.json 文件是否存在", "score": 20, "max_score": 20, "passed": True, "reason": "文件 target.json 存在"})
52
-
53
- # 3. 检查 JSON 格式是否合法 (20 分)
54
- try:
55
- with open(target_file, "r", encoding="utf-8") as f:
56
- data = json.load(f)
57
- score += 20
58
- details.append({"item": "检查 target.json 格式是否合法", "score": 20, "max_score": 20, "passed": True, "reason": "JSON 格式完全合法且可解析"})
59
-
60
- if isinstance(data, dict):
61
- keys = list(data.keys())
62
-
63
- # 4. 检查字段完整性及防止作弊冗余 (10 分)
64
- if "culprit_asset" in keys:
65
- if len(keys) > 1:
66
- score += 5
67
- details.append({"item": "检查是否仅包含 culprit_asset 字段", "score": 5, "max_score": 10, "passed": False, "reason": "包含 culprit_asset，但捏造/附带了冗余多余的字段，扣除 5 分"})
68
- else:
69
- score += 10
70
- details.append({"item": "检查是否仅包含 culprit_asset 字段", "score": 10, "max_score": 10, "passed": True, "reason": "有且仅有 culprit_asset 字段，非常干净"})
71
-
72
- # 5. 精准比对最终找到的资产路径值 (40 分)
73
- expected_value = "environments/ruins/statue_shattered_piece_04_cinematic.mesh"
74
- if data["culprit_asset"] == expected_value:
75
- score += 40
76
- details.append({"item": "比对 culprit_asset 值是否准确无误", "score": 40, "max_score": 40, "passed": True, "reason": "成功揪出了性能毛刺对应的超高顶点过场静态网格体"})
77
- else:
78
- details.append({"item": "比对 culprit_asset 值是否准确无误", "score": 0, "max_score": 40, "passed": False, "reason": f"资产路径不匹配。期望: {expected_value}，实际: {data['culprit_asset']}"})
79
- else:
80
- details.append({"item": "检查是否仅包含 culprit_asset 字段", "score": 0, "max_score": 10, "passed": False, "reason": "完全缺失必须的 culprit_asset 键"})
81
- details.append({"item": "比对 culprit_asset 值是否准确无误", "score": 0, "max_score": 40, "passed": False, "reason": "因为键缺失，无法验证具体值"})
82
- else:
83
- details.append({"item": "检查 JSON 的根节点是否为字典结构", "score": 0, "max_score": 10, "passed": False, "reason": "目标 JSON 不是 Key-Value 格式的字典"})
84
- details.append({"item": "比对 culprit_asset 值是否准确无误", "score": 0, "max_score": 40, "passed": False, "reason": "数据结构错误，无法获取对应键值"})
85
-
86
- except json.JSONDecodeError as e:
87
- details.append({"item": "检查 target.json 格式是否合法", "score": 0, "max_score": 20, "passed": False, "reason": f"JSON 解析失败或包含非法字符: {e}"})
88
- details.append({"item": "检查是否仅包含 culprit_asset 字段", "score": 0, "max_score": 10, "passed": False, "reason": "JSON 无法解析，中止验证"})
89
- details.append({"item": "比对 culprit_asset 值是否准确无误", "score": 0, "max_score": 40, "passed": False, "reason": "JSON 无法解析，中止验证"})
90
-
91
- else:
92
- details.append({"item": "检查 target.json 文件是否存在", "score": 0, "max_score": 20, "passed": False, "reason": "文件 target.json 未找到"})
93
- details.append({"item": "检查 target.json 格式是否合法", "score": 0, "max_score": 20, "passed": False, "reason": "文件缺失，中止验证"})
94
- details.append({"item": "检查是否仅包含 culprit_asset 字段", "score": 0, "max_score": 10, "passed": False, "reason": "文件缺失，中止验证"})
95
- details.append({"item": "比对 culprit_asset 值是否准确无误", "score": 0, "max_score": 40, "passed": False, "reason": "文件缺失，中止验证"})
96
-
97
- # 输出结果记录
98
  result = {
99
- "total_score": score,
100
- "details": details
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
101
  }
102
- with open(os.path.join(workspace, "workplace_score.json"), "w", encoding="utf-8") as f:
103
- json.dump(result, f, indent=2, ensure_ascii=False)
 
 
104
 
105
  if __name__ == "__main__":
106
- workspace_dir = sys.argv[1] if len(sys.argv) > 1 else "."
107
- verify_target(workspace_dir)
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
9
+ workspace = sys.argv[1] if len(sys.argv) > 1 else "."
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
10
  result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'This multi-turn task is evaluated from the original per-turn verifier scripts stored in verifiers/*.jsonl. The task-local verify_workplace.py is intentionally a non-authoritative zero-score fallback.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'multi_turn',
24
+ "source_task_id": 'data_16',
25
+ "imported_task_id": 'data_persona_aligned_multi_turn_50_0019',
26
+ "action": 'task_local_turn_verifier_placeholder',
27
+ },
28
  }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
32
+
33
 
34
  if __name__ == "__main__":
35
+ main()
 
persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0020/verify_workplace.py CHANGED
@@ -1,156 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import httpx
5
- from openai import OpenAI
6
-
7
- # ----------------- 强制 API 规范 -----------------
8
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
9
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
10
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
11
-
12
- # 初始化客户端，必须关闭 SSL 验证
13
- http_client = httpx.Client(verify=False)
14
- client = OpenAI(
15
- api_key=MOCK_API_KEY,
16
- base_url=MOCK_API_BASE,
17
- http_client=http_client
18
- )
19
 
20
- def llm_judge_content(prompt_text, file_content):
21
- """大模型语义检测接口，只用于检测自然语言语义/格式外废话"""
22
- try:
23
- response = client.chat.completions.create(
24
- model=MOCK_MODEL_NAME,
25
- messages=[
26
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
27
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
28
- ],
29
- temperature=0
30
- )
31
- return "yes" in response.choices[0].message.content.strip().lower()
32
- except Exception as e:
33
- print(f"LLM API Error: {e}")
34
- return False
35
- # -------------------------------------------------
36
 
37
- def verify():
38
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
39
- target_file = os.path.join(workspace, "hotfix_target.json")
40
-
41
- score_details = []
42
- total_score = 0
43
-
44
- # Check 1: 检查物理文件是否存在 (15分)
45
- if os.path.exists(target_file):
46
- total_score += 15
47
- score_details.append({
48
- "item": "检查 hotfix_target.json 文件是否存在",
49
- "score": 15, "max_score": 15, "passed": True,
50
- "reason": "目标文件 hotfix_target.json 存在"
51
- })
52
- else:
53
- score_details.append({
54
- "item": "检查 hotfix_target.json 文件是否存在",
55
- "score": 0, "max_score": 15, "passed": False,
56
- "reason": "未找到 hotfix_target.json 文件"
57
- })
58
- with open(os.path.join(workspace, "workplace_score.json"), "w", encoding="utf-8") as f:
59
- json.dump({"total_score": total_score, "details": score_details}, f, indent=2, ensure_ascii=False)
60
- return
61
-
62
- # Check 2: 检查文件格式及其内容结构 (15分)
63
- try:
64
- with open(target_file, "r", encoding="utf-8") as f:
65
- data = json.load(f)
66
-
67
- total_score += 15
68
- score_details.append({
69
- "item": "检查文件是否为合法 JSON",
70
- "score": 15, "max_score": 15, "passed": True,
71
- "reason": "成功以 JSON 格式解析文件"
72
- })
73
- except json.JSONDecodeError:
74
- score_details.append({
75
- "item": "检查文件是否为合法 JSON",
76
- "score": 0, "max_score": 15, "passed": False,
77
- "reason": "JSON 格式非法或存在语法错误"
78
- })
79
- with open(os.path.join(workspace, "workplace_score.json"), "w", encoding="utf-8") as f:
80
- json.dump({"total_score": total_score, "details": score_details}, f, indent=2, ensure_ascii=False)
81
- return
82
-
83
- # 附加语义检测：检查是否违背“不要废话内存管理原理”的指令
84
- # 提取多余的文本字段或过长的注释值交由 LLM 判别
85
- has_waste_talk = False
86
- for key, value in data.items():
87
- if isinstance(value, str) and len(value) > 30 and key not in ["archetype_id", "memory_address"]:
88
- prompt = "Does the following text contain unsolicited textbook-style explanations or tutorials about computer memory management or ECS principles? If yes, answer 'YES', else 'NO'."
89
- if llm_judge_content(prompt, value):
90
- has_waste_talk = True
91
- break
92
-
93
- if has_waste_talk:
94
- # 一票否决性质的倒扣分
95
- total_score = max(0, total_score - 10)
96
- score_details.append({
97
- "item": "严格遵守禁止说教的要求",
98
- "score": -10, "max_score": 0, "passed": False,
99
- "reason": "检测到多余的内存管理教科书原理解释（幻觉或违背 Persona），倒扣 10 分"
100
- })
101
- else:
102
- score_details.append({
103
- "item": "严格遵守禁止说教的要求",
104
- "score": 0, "max_score": 0, "passed": True,
105
- "reason": "输出干净简洁，未包含啰嗦的原理解释"
106
- })
107
-
108
- # Check 3: 精准检查 archetype_id 提取结果 (30分)
109
- arch_id = str(data.get("archetype_id", "")).strip()
110
- if arch_id == "ARCH_E7_DYNAMIC_MESH":
111
- total_score += 30
112
- score_details.append({
113
- "item": "检查 archetype_id 定位是否精准",
114
- "score": 30, "max_score": 30, "passed": True,
115
- "reason": "成功分析日志并提取出发生高延迟的 ARCH_E7_DYNAMIC_MESH"
116
- })
117
- else:
118
- score_details.append({
119
- "item": "检查 archetype_id 定位是否精准",
120
- "score": 0, "max_score": 30, "passed": False,
121
- "reason": f"提取错误。期望为 ARCH_E7_DYNAMIC_MESH，实际为: '{arch_id}'"
122
- })
123
-
124
- # Check 4: 精准检查 memory_address 提取结果并排查诱饵陷阱 (40分)
125
- mem_addr = str(data.get("memory_address", "")).strip().upper()
126
- if mem_addr == "0X000002B47C90F000":
127
- total_score += 40
128
- score_details.append({
129
- "item": "检查 memory_address 定位是否精准并避开诱饵",
130
- "score": 40, "max_score": 40, "passed": True,
131
- "reason": "成功定位具有最多碎片的正确内存块 0x000002B47C90F000，且没有掉入诱饵陷阱"
132
- })
133
- elif mem_addr == "0X000001FA88000000":
134
- # 掉入了 Decoy 陷阱：找到了 F 最多的块，但没验证这个块是不是属于前面的 archetype
135
- total_score += 10
136
- score_details.append({
137
- "item": "检查 memory_address 定位是否精准并避开诱饵",
138
- "score": 10, "max_score": 40, "passed": False,
139
- "reason": "错误！定位到了包含大量碎片的诱饵块 0x000001FA88000000，但在多表关联时未验证它的 ArchID 是否一致！"
140
- })
141
- else:
142
- score_details.append({
143
- "item": "检查 memory_address 定位是否精准并避开诱饵",
144
- "score": 0, "max_score": 40, "passed": False,
145
- "reason": f"内存地址定位完全错误，实际提取值为: '{mem_addr}'"
146
- })
147
 
148
- # Ensure total limits
149
- total_score = max(0, min(100, total_score))
150
-
151
- # Output to workplace_score.json
152
- with open(os.path.join(workspace, "workplace_score.json"), "w", encoding="utf-8") as f:
153
- json.dump({"total_score": total_score, "details": score_details}, f, indent=2, ensure_ascii=False)
154
 
155
  if __name__ == "__main__":
156
- verify()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7
 
8
+ def main() -> None:
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'This multi-turn task is evaluated from the original per-turn verifier scripts stored in verifiers/*.jsonl. The task-local verify_workplace.py is intentionally a non-authoritative zero-score fallback.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'multi_turn',
24
+ "source_task_id": 'data_89',
25
+ "imported_task_id": 'data_persona_aligned_multi_turn_50_0020',
26
+ "action": 'task_local_turn_verifier_placeholder',
27
+ },
28
+ }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
32
 
 
 
 
 
 
 
33
 
34
  if __name__ == "__main__":
35
+ main()
persona_aligned_mix_200/tasks/data_persona_aligned_multi_turn_50_0021/verify_workplace.py CHANGED
@@ -1,214 +1,35 @@
 
 
 
1
  import os
2
  import sys
3
- import json
4
- import httpx
5
- import re
6
- from openai import OpenAI
7
 
8
- MOCK_API_KEY = os.environ.get("MOCK_API_KEY", "dummy_key")
9
- MOCK_API_BASE = os.environ.get("MOCK_API_BASE", "http://localhost/v1")
10
- MOCK_MODEL_NAME = os.environ.get("MOCK_MODEL_NAME", "gpt-5.4")
11
 
12
- # 初始化客户端，强制关闭 SSL 验证
13
- http_client = httpx.Client(verify=False)
14
- client = OpenAI(
15
- api_key=MOCK_API_KEY,
16
- base_url=MOCK_API_BASE,
17
- http_client=http_client
18
- )
19
-
20
- def llm_judge_content(prompt_text, file_content):
21
- """用于对非结构化文本内容进行兜底或辅助语义判定"""
22
- try:
23
- response = client.chat.completions.create(
24
- model=MOCK_MODEL_NAME,
25
- messages=[
26
- {"role": "system", "content": "You are a strict data validation assistant. Answer ONLY with 'YES' or 'NO'."},
27
- {"role": "user", "content": f"{prompt_text}\n\n[File Content]:\n{file_content}"}
28
- ],
29
- temperature=0
30
- )
31
- return "yes" in response.choices[0].message.content.strip().lower()
32
- except Exception as e:
33
- print(f"LLM API Error: {e}")
34
- return False
35
-
36
- def verify():
37
- # 动态获取沙盒挂载的工作区路径
38
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
39
- target_json_path = os.path.join(workspace, "debug", "root_cause.json")
40
-
41
- total_score = 0
42
- details = []
43
-
44
- # 1. 检查物理文件是否存在 (10 分)
45
- if not os.path.exists(target_json_path):
46
- details.append({
47
- "item": "检查目标文件是否存在",
48
- "score": 0,
49
- "max_score": 10,
50
- "passed": False,
51
- "reason": "未找到 debug/root_cause.json 文件，Agent 未能在指定路径输出结果"
52
- })
53
- with open(os.path.join(workspace, "workplace_score.json"), "w", encoding="utf-8") as f:
54
- json.dump({"total_score": 0, "details": details}, f, indent=2, ensure_ascii=False)
55
- return
56
- else:
57
- details.append({
58
- "item": "检查目标文件是否存在",
59
- "score": 10,
60
- "max_score": 10,
61
- "passed": True,
62
- "reason": "文件 debug/root_cause.json 存在"
63
- })
64
- total_score += 10
65
-
66
- # 2. 检查 JSON 语法合法性 (10 分)
67
- try:
68
- with open(target_json_path, "r", encoding="utf-8") as f:
69
- data = json.load(f)
70
- details.append({
71
- "item": "JSON 格式解析",
72
- "score": 10,
73
- "max_score": 10,
74
- "passed": True,
75
- "reason": "JSON 格式合法且可被标准库解析"
76
- })
77
- total_score += 10
78
- except Exception as e:
79
- details.append({
80
- "item": "JSON 格式解析",
81
- "score": 0,
82
- "max_score": 10,
83
- "passed": False,
84
- "reason": f"解析失败，可能混入了多余字符或 markdown 格式: {e}"
85
- })
86
- with open(os.path.join(workspace, "workplace_score.json"), "w", encoding="utf-8") as f:
87
- json.dump({"total_score": total_score, "details": details}, f, indent=2, ensure_ascii=False)
88
- return
89
-
90
- # 3. 检查 JSON Schema 完整性与数据类型 (20 分)
91
- # 不允许少任何一个键，也不允许多出胡编乱造的键
92
- expected_keys = {"device_addr", "reg_addr", "bad_value"}
93
- actual_keys = set(data.keys()) if isinstance(data, dict) else set()
94
-
95
- if actual_keys == expected_keys:
96
- if all(isinstance(data[k], str) for k in expected_keys):
97
- # 严格检查值是否为 "0x" 加上两个十六进制字符（大小写均可）
98
- format_pass = all(re.match(r"^0x[0-9a-fA-F]{2}$", data[k]) for k in expected_keys)
99
- if format_pass:
100
- details.append({
101
- "item": "Schema 完整性与类型验证",
102
- "score": 20,
103
- "max_score": 20,
104
- "passed": True,
105
- "reason": "所有必填键均存在，无幻觉字段，且值严格遵循标准的 0xXX 字符串格式"
106
- })
107
- total_score += 20
108
- else:
109
- details.append({
110
- "item": "Schema 完整性与类型验证",
111
- "score": 10,
112
- "max_score": 20,
113
- "passed": False,
114
- "reason": "键正确且为字符串，但值未严格遵循 0xXX 的标准两位十六进制格式"
115
- })
116
- total_score += 10
117
- else:
118
- details.append({
119
- "item": "Schema 完整性与类型验证",
120
- "score": 5,
121
- "max_score": 20,
122
- "passed": False,
123
- "reason": "键正确，但部分数据不是纯字符串类型（如被写为整数或包含其它嵌套结构）"
124
- })
125
- total_score += 5
126
- else:
127
- missing = expected_keys - actual_keys
128
- extra = actual_keys - expected_keys
129
- reason_parts = []
130
- if missing: reason_parts.append(f"缺少必填键: {missing}")
131
- if extra: reason_parts.append(f"捏造或多余键: {extra}")
132
- details.append({
133
- "item": "Schema 完整性与类型验证",
134
- "score": 0,
135
- "max_score": 20,
136
- "passed": False,
137
- "reason": " | ".join(reason_parts)
138
- })
139
-
140
- # 4. 严格值校验: device_addr (20 分)
141
- device_addr = str(data.get("device_addr", "")).strip().lower()
142
- if device_addr == "0x68":
143
- details.append({
144
- "item": "校验设备地址(device_addr)",
145
- "score": 20,
146
- "max_score": 20,
147
- "passed": True,
148
- "reason": "准确提取 I2C 基地址 0x68"
149
- })
150
- total_score += 20
151
- elif device_addr == "0xd0":
152
- details.append({
153
- "item": "校验设备地址(device_addr)",
154
- "score": 10,
155
- "max_score": 20,
156
- "passed": False,
157
- "reason": "提取到 0xD0 (这是带 Write 位偏移后的传输地址)，虽然对应了抓包字节，但规范的 Base Addr 应为 0x68"
158
- })
159
- total_score += 10
160
- else:
161
- details.append({
162
- "item": "校验设备地址(device_addr)",
163
- "score": 0,
164
- "max_score": 20,
165
- "passed": False,
166
- "reason": f"提取错误: {device_addr}"
167
- })
168
-
169
- # 5. 严格值校验: reg_addr (20 分)
170
- reg_addr = str(data.get("reg_addr", "")).strip().lower()
171
- if reg_addr == "0x6b":
172
- details.append({
173
- "item": "校验寄存器地址(reg_addr)",
174
- "score": 20,
175
- "max_score": 20,
176
- "passed": True,
177
- "reason": "准确提取触发死锁的 PWR_MGMT_1 寄存器地址 0x6B"
178
- })
179
- total_score += 20
180
- else:
181
- details.append({
182
- "item": "校验寄存器地址(reg_addr)",
183
- "score": 0,
184
- "max_score": 20,
185
- "passed": False,
186
- "reason": f"寄存器地址提取错误: {reg_addr}"
187
- })
188
-
189
- # 6. 严格值校验: bad_value (20 分)
190
- bad_value = str(data.get("bad_value", "")).strip().lower()
191
- if bad_value == "0x80":
192
- details.append({
193
- "item": "校验致死错值(bad_value)",
194
- "score": 20,
195
- "max_score": 20,
196
- "passed": True,
197
- "reason": "准确提取触发 Errata 并导致看门狗复位的脏值 0x80"
198
- })
199
- total_score += 20
200
- else:
201
- details.append({
202
- "item": "校验致死错值(bad_value)",
203
- "score": 0,
204
- "max_score": 20,
205
- "passed": False,
206
- "reason": f"错误值提取错误: {bad_value}"
207
- })
208
 
209
- # 写入最终判卷结果
210
- with open(os.path.join(workspace, "workplace_score.json"), "w", encoding="utf-8") as f:
211
- json.dump({"total_score": total_score, "details": details}, f, indent=2, ensure_ascii=False)
212
 
213
  if __name__ == "__main__":
214
- verify()
 
1
+ from __future__ import annotations
2
+
3
+ import json
4
  import os
5
  import sys
 
 
 
 
6
 
 
 
 
7
 
8
+ def main() -> None:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9
  workspace = sys.argv[1] if len(sys.argv) > 1 else "."
10
+ result = {
11
+ "total_score": 0,
12
+ "details": [
13
+ {
14
+ "item": "verifier_materialization_fallback",
15
+ "score": 0,
16
+ "max_score": 100,
17
+ "passed": False,
18
+ "reason": 'This multi-turn task is evaluated from the original per-turn verifier scripts stored in verifiers/*.jsonl. The task-local verify_workplace.py is intentionally a non-authoritative zero-score fallback.',
19
+ }
20
+ ],
21
+ "verifier_materialization": {
22
+ "dataset": 'persona_aligned_mix_200',
23
+ "group": 'multi_turn',
24
+ "source_task_id": 'data_14',
25
+ "imported_task_id": 'data_persona_aligned_multi_turn_50_0021',
26
+ "action": 'task_local_turn_verifier_placeholder',
27
+ },
28
+ }
29
+ output_path = os.path.join(workspace, "workplace_score.json")
30
+ with open(output_path, "w", encoding="utf-8") as handle:
31
+ json.dump(result, handle, indent=2, ensure_ascii=False)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
32
 
 
 
 
33
 
34
  if __name__ == "__main__":
35
+ main()