跳到正文
世傲命的技术观测站 PERSONAL TECHNOLOGY OBSERVATORY

Open data · reviewed public projection

Pilot B:24 次运行的数据浏览器

这里公开实验设计和逐运行数据,方便读者检查每一条结论。页面展示的是经过字段白名单与隐私检查的公开投影,不包含 API Key、私有响应正文、本机路径或内部产物。

实验矩阵
3 × 4 × 2任务类型 × 实例 × 架构
正式运行
24无静默替换运行
模型
deepseek-v4-flashthinking 关闭
通过代码验收
8/12 · 8/12 开放式 · 目标图混合

01 · Study design

先看控制变量,再看结果

两边使用同一模型、同一任务实例、同一组工具、同一代码验收和同一总预算。改变的是模型外面的执行控制方式。
A

开放式执行

Clear tool contract, all tools visible, complete trace; the model chooses planning, recovery, and stop.

B

目标图混合执行

Explicit goal/dependency state, all currently legal frontier actions visible, evidence ledger, one-attempt failed-route recovery, and deterministic bounded parallel branches for independent decomposition goals.

预算

每条运行统一计费

  • 模型调用 ≤ 18
  • 工具调用 ≤ 24
  • 总 tokens ≤ 30,000
  • 并行分支 ≤ 3
  • 超时 240 秒
3 类任务 × 4 个实例 × 2 种架构 = 24 条正式运行
任务类型冻结实例开放式执行目标图混合
替代路径 alternative_path 4 4 4
开放取证 open_evidence 4 4 4
并行拆解 parallel_decomposition 4 4 4

02 · Run explorer

逐条检查 24 次运行

先用筛选缩小范围,再展开单条记录。页面只展示公开指标、结束原因、代码验收结果和响应哈希;完整公开字段可下载 JSON。

显示 24 / 24 条

#实例架构验收效用模型 / 工具tokens时间公开详情
01 替代路径 alternative_path-01 开放式执行 通过 100 5 / 5 2,796 5.02 s
展开

运行与结束

run key
contracted_react__alternative_path-01
termination
finish
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
0
合并冲突
0

执行指标

重复动作
0
无关动作
0
路径切换
关键路径轮次
4
控制器拒绝
0
prompt / completion
2,659 / 137

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · controller P 72acd4fb9649938f93bb1039b5fab440ea9d816649bdfefd7ef37b0a7f544477 R 1199fd234b315946d0650fb28abc53628005ff2fa289c4986b1b55c67fc6af9b
  2. round 2 · controller P 6e5eda99c15c3943a5c6d4ea3d4580d2d0b1250a31c3d0514cbc5dd593e362f8 R 295fdef966214727a1013e96c0eb72f6bbcc1f5d8d5dfb5e6face00863de3983
  3. round 3 · controller P 2f5824f5fb54f30a060a37bec2ec9f902952a30a3a767452a22851444e780fdf R 5ba84ad90f55e4808ccf8e4c18893f24557b94e3851892cf5f07b67f85eb50fe
  4. round 4 · controller P d44a2169948be779c105d862fe2fdce2b565e2594d0069f9d2a5011f09b81dfa R 203d8619b22a85be1124e4a141d0f478c94c94d64b6b9394e7a0c03b6a3c2a92
  5. round 5 · controller P 626bf4aab1bb57f09f70d0706cd7dcaf7d445f66133779b5739a09427acfad97 R 7310372e895dfcd49057382944f726a4b0ab03cbdea6904b834b3516d3bead52
02 替代路径 alternative_path-01 目标图混合执行 通过 100 8 / 5 4,655 8.09 s
展开

运行与结束

run key
goal_graph_hybrid__alternative_path-01
termination
finish
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
0
合并冲突
0

执行指标

重复动作
3
无关动作
0
路径切换
关键路径轮次
4
控制器拒绝
3
prompt / completion
4,375 / 280

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · controller P 04ff39e755b40b22cc16c72b931a6235f5502c9b8c30f0a9f957f1fe0b9f4570 R 1199fd234b315946d0650fb28abc53628005ff2fa289c4986b1b55c67fc6af9b
  2. round 2 · controller P ea9c33161ec96aacb0634bf79ddd27192366af53eeeefc82dfc967f1229bfce9 R 295fdef966214727a1013e96c0eb72f6bbcc1f5d8d5dfb5e6face00863de3983
  3. round 3 · controller P 94034cf43a7ca8a184ae11cf581675f20d6cd3b738dd120f3ee790b2eb133127 R 5ba84ad90f55e4808ccf8e4c18893f24557b94e3851892cf5f07b67f85eb50fe
  4. round 4 · controller P 6a3628b53697f3b93ef65920ad9fe2a70b7a6e0d5ee2828239f0efb499463a9b R 203d8619b22a85be1124e4a141d0f478c94c94d64b6b9394e7a0c03b6a3c2a92
  5. round 5 · controller P 19d242e07a398be1313d1392d30f86035bb8e730d3cd6d81c99391b1b02bf26f R 203d8619b22a85be1124e4a141d0f478c94c94d64b6b9394e7a0c03b6a3c2a92
  6. round 6 · controller P 46522d73c8751dc91c25db3d3d298efe36b3b137a8494f3f087cf6be52307e86 R 203d8619b22a85be1124e4a141d0f478c94c94d64b6b9394e7a0c03b6a3c2a92
  7. round 7 · controller P 304ad45af8ded4b09f25a056cba7fc270002ebd50701ec5470c7f237d531e304 R 843caf795919aa6a0c0052ce9f76769f21efcf847577f788f994d3f6ee5db258
  8. round 8 · controller P 36debcc9dccbd6ae71b2fdb1b1ccfdcba6ce9f137a888462efbd12b8667bbd2a R 7310372e895dfcd49057382944f726a4b0ab03cbdea6904b834b3516d3bead52
03 替代路径 alternative_path-02 目标图混合执行 通过 100 7 / 5 3,985 5.94 s
展开

运行与结束

run key
goal_graph_hybrid__alternative_path-02
termination
finish
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
0
合并冲突
0

执行指标

重复动作
2
无关动作
0
路径切换
关键路径轮次
4
控制器拒绝
2
prompt / completion
3,759 / 226

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · controller P 2115969a0ba27437d5143c3276f4352d329a4d6c776f689508425e8fa86f6817 R 1199fd234b315946d0650fb28abc53628005ff2fa289c4986b1b55c67fc6af9b
  2. round 2 · controller P 2c676b4112e9e04a062c7cc7afc1355dae8c6ef7c329516de17d382cc9d32788 R 79566c274ead8a167911b979d693844cdcca5cf0857b42da9b1eb239a88db31d
  3. round 3 · controller P 17c2dee98c19a71502f542b9181cf47601a62fa25c5c2d81d9348b7708597246 R c33d722124abdf6782d7cf08b98e2f96db8f28c6245a9e15b92d9ce08f191b5d
  4. round 4 · controller P f6e8b66c14ad4728ba945801e48f15754f73ae30d32f29458c453bc9ac250cd7 R cd1223a2de4b951b7b1d4b63c6e5c041fae13c5cfb5035f9e6a423bfa7cd8b09
  5. round 5 · controller P 1319510d1c9a301d9e23b80ea3c6503b4ef3f8b0502a7f3cfe2d0feeb24d2755 R cd1223a2de4b951b7b1d4b63c6e5c041fae13c5cfb5035f9e6a423bfa7cd8b09
  6. round 6 · controller P f3765159a32d3e5f23e8f8454b32a3e69aa4d77538f2bc82b31520096fed9105 R cd1223a2de4b951b7b1d4b63c6e5c041fae13c5cfb5035f9e6a423bfa7cd8b09
  7. round 7 · controller P 2718f24ad7196b5d00f223ba853a833e4e946b976db12ae66d493991b9eb3703 R 7310372e895dfcd49057382944f726a4b0ab03cbdea6904b834b3516d3bead52
04 替代路径 alternative_path-02 开放式执行 通过 100 5 / 5 2,784 3.93 s
展开

运行与结束

run key
contracted_react__alternative_path-02
termination
finish
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
0
合并冲突
0

执行指标

重复动作
0
无关动作
0
路径切换
关键路径轮次
4
控制器拒绝
0
prompt / completion
2,650 / 134

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · controller P f272a0d5bee5925608bbea495711358613818d8f366f9aa33924cbd286eefe39 R 1199fd234b315946d0650fb28abc53628005ff2fa289c4986b1b55c67fc6af9b
  2. round 2 · controller P 040d50bf9f9f15b3928fdf1aac5a34cec068c983ad3f5f94d19f17170f8b8717 R 79566c274ead8a167911b979d693844cdcca5cf0857b42da9b1eb239a88db31d
  3. round 3 · controller P ab5cf09aece3c7dd340332af5f61bf38a3e96c3d89f98ef15b072186dba0512c R c33d722124abdf6782d7cf08b98e2f96db8f28c6245a9e15b92d9ce08f191b5d
  4. round 4 · controller P 289c24ecef9be0dbb55f60068d7e41ab1b5916ac5c6efa1521025bd4be0f4978 R cd1223a2de4b951b7b1d4b63c6e5c041fae13c5cfb5035f9e6a423bfa7cd8b09
  5. round 5 · controller P a87e48926f99e70cbd2f40ac90f902be43296758a3cf783b71a6c9ca9d4fe912 R 7310372e895dfcd49057382944f726a4b0ab03cbdea6904b834b3516d3bead52
05 替代路径 alternative_path-03 开放式执行 通过 100 5 / 5 2,796 4.80 s
展开

运行与结束

run key
contracted_react__alternative_path-03
termination
finish
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
0
合并冲突
0

执行指标

重复动作
0
无关动作
0
路径切换
关键路径轮次
4
控制器拒绝
0
prompt / completion
2,659 / 137

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · controller P 6933acd555be72210f8aaeab2f16cf20c7ae699a6ee05ca3d181ae109d183748 R 1199fd234b315946d0650fb28abc53628005ff2fa289c4986b1b55c67fc6af9b
  2. round 2 · controller P 77386662a941f700b893545cf5cb91ab296fe2d6c31f01bccd661099a98d47f8 R d6786857b1d395171819efc3597fcc3a65bd4036e5156a4bcfcb7ab5e5abd27e
  3. round 3 · controller P 5280c28487225d77892cdde5653a5ce751cbc8f369ddbd9f143f29ee80abf398 R 4acb41bc8505adf99b0077525f9b09033bc4edd0c91a9f7418e91f49b6bfac0c
  4. round 4 · controller P ea04f365489ae3c444eb4d012da43dd8d0d3c8f74b26dff3b69dabcd0b0098b9 R 21e98c2581232bb2a7ff2e949420bb28a8d000ce06669eff32173fe7f1ac716b
  5. round 5 · controller P f3a0b6095f6d33484ee99872954aa12532b319706ae6d0e2832f566b23406277 R 7310372e895dfcd49057382944f726a4b0ab03cbdea6904b834b3516d3bead52
06 替代路径 alternative_path-03 目标图混合执行 通过 100 10 / 5 6,020 8.86 s
展开

运行与结束

run key
goal_graph_hybrid__alternative_path-03
termination
finish
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
0
合并冲突
0

执行指标

重复动作
5
无关动作
0
路径切换
关键路径轮次
4
控制器拒绝
5
prompt / completion
5,649 / 371

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · controller P ba33ac721d2db2e6b853fe352db9e12017210a6ffaad43af35b0866e31e95814 R 1199fd234b315946d0650fb28abc53628005ff2fa289c4986b1b55c67fc6af9b
  2. round 2 · controller P 2337632e6731426dd0d518f70c690fba8d0f6895301c62822968ab4bbf415c44 R d6786857b1d395171819efc3597fcc3a65bd4036e5156a4bcfcb7ab5e5abd27e
  3. round 3 · controller P cb1d57124859af72b85af30fe27958db2be64c261f087f72801d5f66d7153914 R 217ec23fa932e61e25163a60c1260d0ae0b9133ba664d4c8969a0b58701e4a42
  4. round 4 · controller P 2eb10a7dfa39deb82a348dfaf5746066809855a0faf8cdf06b68f2d36a0dac20 R 21e98c2581232bb2a7ff2e949420bb28a8d000ce06669eff32173fe7f1ac716b
  5. round 5 · controller P 34c751f92c0705f3268495be3b1ab2162fef67508b51073c4204912e67af35b8 R 21e98c2581232bb2a7ff2e949420bb28a8d000ce06669eff32173fe7f1ac716b
  6. round 6 · controller P e035d99f5ee93b3b5246933765e5b148ac348155bafe1e2489c2bf701000c6ef R 21e98c2581232bb2a7ff2e949420bb28a8d000ce06669eff32173fe7f1ac716b
  7. round 7 · controller P 455223ce4b53701cc8b08475c1e4bf08070401aef00e430a9e72b020f3d760be R 21e98c2581232bb2a7ff2e949420bb28a8d000ce06669eff32173fe7f1ac716b
  8. round 8 · controller P 5ed508d7d9880089a57eff25c82243b27ac2f7a5bb058ba1b81bc5b2e06f6338 R 21e98c2581232bb2a7ff2e949420bb28a8d000ce06669eff32173fe7f1ac716b
  9. round 9 · controller P bce5c741bb08c5faadba32952a3f8563b7694a057e2f2cd8e34c980ebba2191a R 21e98c2581232bb2a7ff2e949420bb28a8d000ce06669eff32173fe7f1ac716b
  10. round 10 · controller P ba8adcd7371952ec126107ae7a7e22768190a07540de10af8b9a49755c277f12 R 7310372e895dfcd49057382944f726a4b0ab03cbdea6904b834b3516d3bead52
07 替代路径 alternative_path-04 目标图混合执行 通过 100 18 / 4 12,403 16.21 s
展开

运行与结束

run key
goal_graph_hybrid__alternative_path-04
termination
model_budget_exhausted
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
0
合并冲突
0

执行指标

重复动作
14
无关动作
0
路径切换
关键路径轮次
4
控制器拒绝
14
prompt / completion
11,610 / 793

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · controller P fe423a66de39086636aef39a075e3a5b03e7a48879f2b61022b247458b5e1643 R 1199fd234b315946d0650fb28abc53628005ff2fa289c4986b1b55c67fc6af9b
  2. round 2 · controller P 0234e76d1cf8c59fe7c1814463a48a63bdd15da972ad204ec4ce271133143083 R cffe4d0d4becfa93b65dea9bcd03ff111771165b0bd2fc86282ee2f91a8255d3
  3. round 3 · controller P ecc9d41d26c96b2e0a4bb2b6649d7a58353c6e6cd4c3b35147d0c8f969ae3ec5 R f9214d7bd9a1c6c64ee9c88d1ec71bdffcbef343b061238d995a46cd54384324
  4. round 4 · controller P 548393e9c2b831ef856a06fcc13686e04776d19de2c1b038c06c109ede61d972 R 5390266cfd85c4e0f60544f0b518936e815e99d160e573344981cbe476d51e58
  5. round 5 · controller P f6368ad974d9586d31d7e9ad88fb4d6a632fd0483f13095c411451d15a273f7a R 5390266cfd85c4e0f60544f0b518936e815e99d160e573344981cbe476d51e58
  6. round 6 · controller P 21eb2bf4907e90834e637fe890cb3a5427f506472a4307535ed4ea0288bcb4b0 R 5390266cfd85c4e0f60544f0b518936e815e99d160e573344981cbe476d51e58
  7. round 7 · controller P 6f9d50d6c1e3dc2930c6625ff43716ecc312855d821e417df42f5b9e61e42490 R 5390266cfd85c4e0f60544f0b518936e815e99d160e573344981cbe476d51e58
  8. round 8 · controller P 071201a7a3744702a05b0854f4659da0e789b66ced9c38df780982d4d68ff403 R 5390266cfd85c4e0f60544f0b518936e815e99d160e573344981cbe476d51e58
  9. round 9 · controller P 4f81503f48f304c5adef12667d22ac99c34d7eea8dc3f2c9f860686c859c4280 R 5390266cfd85c4e0f60544f0b518936e815e99d160e573344981cbe476d51e58
  10. round 10 · controller P 2f3576da857b68fde18cc22859274fc689124f1a90938f9c144fbda828cd9ca9 R 5390266cfd85c4e0f60544f0b518936e815e99d160e573344981cbe476d51e58
  11. round 11 · controller P eaf99452b4fe7f572207b3d184840559630c0f27bc3b1b235a8087f8beded86b R 5390266cfd85c4e0f60544f0b518936e815e99d160e573344981cbe476d51e58
  12. round 12 · controller P a05f45bd2ba427e42cba6e19e5bef7152088d587f7143b32573b623f44d66903 R 5390266cfd85c4e0f60544f0b518936e815e99d160e573344981cbe476d51e58
  13. round 13 · controller P 6cd2e3865584cebe6f9ba54a05f2fce0ada0dd36f1eac0d8f89785d43cb1beb8 R 5390266cfd85c4e0f60544f0b518936e815e99d160e573344981cbe476d51e58
  14. round 14 · controller P a25aa6b822f561227d48bc64e5e066c93e1abf2103784abc13c682298403d5db R 5390266cfd85c4e0f60544f0b518936e815e99d160e573344981cbe476d51e58
  15. round 15 · controller P 1c3e7940f099911f00d67c88c88a12d8137094e010ed8d6313badcd53836c8c9 R 5390266cfd85c4e0f60544f0b518936e815e99d160e573344981cbe476d51e58
  16. round 16 · controller P 0112f1e46817e935424f93245eee16cc21151e837e2cad9af45a83117f348bd2 R 5390266cfd85c4e0f60544f0b518936e815e99d160e573344981cbe476d51e58
  17. round 17 · controller P 6159e7cbde89e41cfceed59d8acb63cd84f974164734d94b1ed1d689a2c26b56 R 5390266cfd85c4e0f60544f0b518936e815e99d160e573344981cbe476d51e58
  18. round 18 · controller P 3f8bb64a8a00c8dd13b2ae690723b420f474d28fead3a461c90d4f69c692484b R 5390266cfd85c4e0f60544f0b518936e815e99d160e573344981cbe476d51e58
08 替代路径 alternative_path-04 开放式执行 通过 100 5 / 5 2,792 4.01 s
展开

运行与结束

run key
contracted_react__alternative_path-04
termination
finish
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
0
合并冲突
0

执行指标

重复动作
0
无关动作
0
路径切换
关键路径轮次
4
控制器拒绝
0
prompt / completion
2,656 / 136

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · controller P 5ef9c5922aa3e1a9bf1cf13714c5adbe83d6ed0b3b2456327ea83a64d6b2d674 R 1199fd234b315946d0650fb28abc53628005ff2fa289c4986b1b55c67fc6af9b
  2. round 2 · controller P 3491edefbf627272a01620542399d833ed23ebee5f5ae4549f94669d86d9fe9e R cffe4d0d4becfa93b65dea9bcd03ff111771165b0bd2fc86282ee2f91a8255d3
  3. round 3 · controller P b11b2ec9a1bc841d6c2a918a0f9176d5ed0d8673061dbd4be95ad51b6ba5de8c R f9214d7bd9a1c6c64ee9c88d1ec71bdffcbef343b061238d995a46cd54384324
  4. round 4 · controller P a5163d00dc786a238672d52c6e173db250726b5c8a81df5ea2730a626f73a046 R 5390266cfd85c4e0f60544f0b518936e815e99d160e573344981cbe476d51e58
  5. round 5 · controller P c81fa0bd2981c65a22149e5f24096161e30648171ece638733198491a747af88 R 7310372e895dfcd49057382944f726a4b0ab03cbdea6904b834b3516d3bead52
09 开放取证 open_evidence-01 开放式执行 通过 100 6 / 6 4,318 5.13 s
展开

运行与结束

run key
contracted_react__open_evidence-01
termination
finish
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
0
合并冲突
0

执行指标

重复动作
0
无关动作
0
路径切换
关键路径轮次
5
控制器拒绝
0
prompt / completion
4,120 / 198

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · controller P 091caacb8265f87c1ad6255d34d58c205183e143fd550de3bc9a56f3a0219bc7 R 1199fd234b315946d0650fb28abc53628005ff2fa289c4986b1b55c67fc6af9b
  2. round 2 · controller P 46a0e0876175c6456f3d5aa22ac0e2e6fabe752cc63c72b0cabcc6879f967cce R 83974279ee363b0f44ba453803a6840d09132b3dfa433f960fb6edb1fb9d1136
  3. round 3 · controller P e347de150e6b1433bc12a3471321d0c595250cc17cfe3fff947d782b7340441d R f78bf6fcf38377ff9cf4bcaf2b8bedc28bec540921b67a059fc3ba69b4c795c8
  4. round 4 · controller P 97bdc1f020654e706790eb3d7a2b5c0be98eba723f621051f98d8f4f045d902d R 766478af48faa1ae324506451d32fdb13724e9f2a6864bbc04c0b06c847044e5
  5. round 5 · controller P 7df8cb58a8b57737018663dc78c1111d3f66bb544ebfb97f6b20a3e5a71b967e R 16a25e35fee767802b2919d79136ea77f5c2a258684d02865f452c6c1524b4c0
  6. round 6 · controller P 027a9948bc29cfaf805f2b4e43816003549e489124bb235f61a2ecd4a803de37 R 7310372e895dfcd49057382944f726a4b0ab03cbdea6904b834b3516d3bead52
10 开放取证 open_evidence-01 目标图混合执行 通过 100 18 / 5 15,220 18.77 s
展开

运行与结束

run key
goal_graph_hybrid__open_evidence-01
termination
model_budget_exhausted
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
0
合并冲突
0

执行指标

重复动作
13
无关动作
0
路径切换
关键路径轮次
5
控制器拒绝
13
prompt / completion
13,989 / 1,231

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · controller P ad0fc9e5a3ebd3522e0e8e3bd4b4ed5d14f9252c9c382650bab6ba698809ca7b R 0929f8a7ea7640da08e12673da0f1f33c8572bd3cf1f7721e9a0e0984916ada1
  2. round 2 · controller P 8060f21995bbda73bfd4f889ba7bef9ecb1a3f300beec5ebcd05408ec5646cca R 83974279ee363b0f44ba453803a6840d09132b3dfa433f960fb6edb1fb9d1136
  3. round 3 · controller P 2fd7e16a27a83c2ea6e7a083c72cc32a476fb1f446f721f90348adccfd86fe99 R f78bf6fcf38377ff9cf4bcaf2b8bedc28bec540921b67a059fc3ba69b4c795c8
  4. round 4 · controller P 25cb5e66517377947084381d98e5d32f4f441016ed37a112e3bd2811c7405a23 R 766478af48faa1ae324506451d32fdb13724e9f2a6864bbc04c0b06c847044e5
  5. round 5 · controller P e18e4c3a0116cf57cc3d8adafcc0eead94f4e4c71b8f2f10b7752f51fb386433 R 16a25e35fee767802b2919d79136ea77f5c2a258684d02865f452c6c1524b4c0
  6. round 6 · controller P 8d57bd8b3def283739d1557d88db847172da131dc97841243c8d008d28721769 R 16a25e35fee767802b2919d79136ea77f5c2a258684d02865f452c6c1524b4c0
  7. round 7 · controller P 1dc7e8a22bd891dd67f8b1042c50a7ad02f02e2d713a4d724b9871d660481d4c R 16a25e35fee767802b2919d79136ea77f5c2a258684d02865f452c6c1524b4c0
  8. round 8 · controller P bc972f4b612b5e013d4171f728c8e79cddbfdcb9c2b292f3a4c86c7e2310e31b R 16a25e35fee767802b2919d79136ea77f5c2a258684d02865f452c6c1524b4c0
  9. round 9 · controller P 550171a5e4a779a098586d2175d31635e392156f955ae31ba16fd34c1f0ee0b4 R 16a25e35fee767802b2919d79136ea77f5c2a258684d02865f452c6c1524b4c0
  10. round 10 · controller P ffe17e307004747b750d42005809feb2b0c5383787f55846096c8d02f978cf5a R 16a25e35fee767802b2919d79136ea77f5c2a258684d02865f452c6c1524b4c0
  11. round 11 · controller P 3f3a9409271867981c6a5900fc7c5bdcf694b0ae5a2151b32c89102d5f8fd5e5 R 16a25e35fee767802b2919d79136ea77f5c2a258684d02865f452c6c1524b4c0
  12. round 12 · controller P 3a9943dcc3ce4176519f6c501c8b7275f0ad09b2fc08520c9e3286f97aaf7c03 R 16a25e35fee767802b2919d79136ea77f5c2a258684d02865f452c6c1524b4c0
  13. round 13 · controller P 839bf2e98d3182bcd91ebd9d87dbc7cf16889bf66aced480ce9dd6d46c5610e2 R 16a25e35fee767802b2919d79136ea77f5c2a258684d02865f452c6c1524b4c0
  14. round 14 · controller P f77b32c25754d21fef9793b7c4d9f7ca6de867756c5f459e82b6ee42c524057f R 16a25e35fee767802b2919d79136ea77f5c2a258684d02865f452c6c1524b4c0
  15. round 15 · controller P e4e9d6e710e073f9235424b07c9158c3be6bd6bd185ef1c0ad5a653fdef01a93 R 16a25e35fee767802b2919d79136ea77f5c2a258684d02865f452c6c1524b4c0
  16. round 16 · controller P 20f7b6be9ac7cef3ee9c9c837f83a4077b1cb094e9307d5f35f26832bc448198 R 16a25e35fee767802b2919d79136ea77f5c2a258684d02865f452c6c1524b4c0
  17. round 17 · controller P e33f1f9ad0dafcd7249e0a8b1ea187fe08dcc989a9c16fc44b05a34c3a20f976 R 16a25e35fee767802b2919d79136ea77f5c2a258684d02865f452c6c1524b4c0
  18. round 18 · controller P 29fdced443a60758a23e8427ca23f2e1b9f5dad9eb2725e78b528bc6577dc374 R 16a25e35fee767802b2919d79136ea77f5c2a258684d02865f452c6c1524b4c0
11 开放取证 open_evidence-02 目标图混合执行 通过 100 18 / 5 15,225 20.74 s
展开

运行与结束

run key
goal_graph_hybrid__open_evidence-02
termination
model_budget_exhausted
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
0
合并冲突
0

执行指标

重复动作
13
无关动作
0
路径切换
关键路径轮次
5
控制器拒绝
13
prompt / completion
13,989 / 1,236

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · controller P a45089fbc9271922275e2617e6846724c65946791a5cec81951263b60c7f3cf7 R 1199fd234b315946d0650fb28abc53628005ff2fa289c4986b1b55c67fc6af9b
  2. round 2 · controller P d818f9806a6758e71442cc033eb0275031cdc69f940d7c15f25c46f7cd0ed033 R ee437febbd1cd556a3212ed0c24af62e7ee29f24289f08347b9c98358c5ec3e1
  3. round 3 · controller P 6970646925b899c02272d875868b99846f358dc82768d08c3f81f46a3b18b52e R 87d1decd1bc4ee0f979ffbc691561d94d7ed52c43a16aa8bd702dde00bca960e
  4. round 4 · controller P 6076ced79fe7348a5affcf0e238cc3166c971146ddf40a5863145bc15327cc0b R 59aad9050a4f0b3b02fc8dcd20c956117dd8953d1b75b057bdcfbe620e1f389c
  5. round 5 · controller P a3e3b4abd45dbfba13dcc1e610ccd837e4eb3cbba99f6f020251fb5f3489a111 R 484187a8612af504caac0b266b76a9263ec394621237cb45bc25ac9cf35f194f
  6. round 6 · controller P 18f585a3fd9647796d13223b99be0a0c637e2dc60a44f6b4d0cf618589a38b29 R 484187a8612af504caac0b266b76a9263ec394621237cb45bc25ac9cf35f194f
  7. round 7 · controller P 12481bd8eeff6b1ed537dc3f734e022d17d9d94efd5fe73a79de254abd447217 R 484187a8612af504caac0b266b76a9263ec394621237cb45bc25ac9cf35f194f
  8. round 8 · controller P feaf58ab521cc714229ac3d85ad2e7da56d21074d77ecd64a2bee9d4b18851b8 R 484187a8612af504caac0b266b76a9263ec394621237cb45bc25ac9cf35f194f
  9. round 9 · controller P 6ee1ab0d0af795bc7568a808d525392f637c4fc3c640b88927490e947b8baafc R 484187a8612af504caac0b266b76a9263ec394621237cb45bc25ac9cf35f194f
  10. round 10 · controller P e86160e1c09e907dbf86c79aa99cfe9eb005d69d90c3393945bf3638609bbe57 R 484187a8612af504caac0b266b76a9263ec394621237cb45bc25ac9cf35f194f
  11. round 11 · controller P 9653a270f855c3e4f59c403b7b22a860b44099845a477fa2ba2f9769206fe14f R 484187a8612af504caac0b266b76a9263ec394621237cb45bc25ac9cf35f194f
  12. round 12 · controller P e8f4d3aeffcc959a00247ba8c856d7cfea56aaf1f7a6ff7bc569f6c4cedb51c0 R 484187a8612af504caac0b266b76a9263ec394621237cb45bc25ac9cf35f194f
  13. round 13 · controller P 375c0d8f459dde9145bcd1333700b029854c7470694295661bdcf0f9f5ac09d7 R 484187a8612af504caac0b266b76a9263ec394621237cb45bc25ac9cf35f194f
  14. round 14 · controller P f11daf369ba86bc71979163be056149315f5bda99b8eb59edb9a3575592e34a8 R 484187a8612af504caac0b266b76a9263ec394621237cb45bc25ac9cf35f194f
  15. round 15 · controller P 5854065d2fe67c0ff9b206ec1d5c2d3a027473eaaf262392c8564d1f85e81b59 R 484187a8612af504caac0b266b76a9263ec394621237cb45bc25ac9cf35f194f
  16. round 16 · controller P 0b3c881330d4ca51dded41c873e74531df511284286ac57e9f8bd6cbcde65c44 R 484187a8612af504caac0b266b76a9263ec394621237cb45bc25ac9cf35f194f
  17. round 17 · controller P 2edde1cbefbd14bf1d8ed08ae30af1991e52548f2bb161fd7cf59839f6a6d209 R 484187a8612af504caac0b266b76a9263ec394621237cb45bc25ac9cf35f194f
  18. round 18 · controller P 9a89babc4880755d5afbc76930aa4c6457ad6023cf61ddb02053d294d1a53957 R 484187a8612af504caac0b266b76a9263ec394621237cb45bc25ac9cf35f194f
12 开放取证 open_evidence-02 开放式执行 通过 100 6 / 6 4,318 6.26 s
展开

运行与结束

run key
contracted_react__open_evidence-02
termination
finish
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
0
合并冲突
0

执行指标

重复动作
0
无关动作
0
路径切换
关键路径轮次
5
控制器拒绝
0
prompt / completion
4,120 / 198

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · controller P 7fc427c153786e310fd3b2a24349d4679c9609a45f12bb34efab01c775cfa526 R 1199fd234b315946d0650fb28abc53628005ff2fa289c4986b1b55c67fc6af9b
  2. round 2 · controller P 601e4de38561a67db3bbbc41fb44fea3c97a05c3363287eada55627e245ef33e R ee437febbd1cd556a3212ed0c24af62e7ee29f24289f08347b9c98358c5ec3e1
  3. round 3 · controller P 32b806600232c15be3739bcc1cc4f22e1ddd2bd41ffa8dd33036201ac65102ad R 87d1decd1bc4ee0f979ffbc691561d94d7ed52c43a16aa8bd702dde00bca960e
  4. round 4 · controller P 10952265add05e7c067f1366131a4c079f1cbce4b1fae8d3d705895c89b862ec R 59aad9050a4f0b3b02fc8dcd20c956117dd8953d1b75b057bdcfbe620e1f389c
  5. round 5 · controller P 37ad50d47a173da07c526798c946c744d6a38fb53acd43c34cc75e6456990f69 R 484187a8612af504caac0b266b76a9263ec394621237cb45bc25ac9cf35f194f
  6. round 6 · controller P 3191ff756143c4fcdde3f28794bb4e593a16a0e9fa3ac3eb13f61dd69266d60c R 7310372e895dfcd49057382944f726a4b0ab03cbdea6904b834b3516d3bead52
13 开放取证 open_evidence-03 开放式执行 通过 100 6 / 6 4,312 5.46 s
展开

运行与结束

run key
contracted_react__open_evidence-03
termination
finish
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
0
合并冲突
0

执行指标

重复动作
0
无关动作
0
路径切换
关键路径轮次
5
控制器拒绝
0
prompt / completion
4,115 / 197

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · controller P 94c1a1828cbb2cf5cc73988a64281ec9d66cdd6cca5400a2b2efcb8275a281ef R 1199fd234b315946d0650fb28abc53628005ff2fa289c4986b1b55c67fc6af9b
  2. round 2 · controller P 48d8997fc911345bd8ec5f727918c6b8624fbb742ca71f6e994fa6eff81c8881 R 6585026d6b9f78fd25ad14cc61a2819bb6e8c63a13c822b5ea071044fbb9f36c
  3. round 3 · controller P 5cdbb0602729105060c075b9c2fdab823d86995cc7836908fb22d3438b08f614 R 10415679d83912250eaf38bdf50b94ee83b78932b4e178945acdd71f57da1c1e
  4. round 4 · controller P 72f2675d4b1ac59e68b116043910ae700cc85790222b5c489edbec14d812dd07 R d4d6c7854a49625d9f6289317ef5fd916199cef00e953b7b32b8a65fe956f224
  5. round 5 · controller P 8c66ba91f11a3e219ab2c8de9a2dd8e382274d63b8a5472ebe00793d5649a27c R e4a64fff85f2685a276ebd85d815426187214d6b3a0f4fe482ba7e252e077faf
  6. round 6 · controller P 8d72143031ea1fe559c8eb584bdf6bd044293134522b17ded6c1006d313d559d R 7310372e895dfcd49057382944f726a4b0ab03cbdea6904b834b3516d3bead52
14 开放取证 open_evidence-03 目标图混合执行 通过 100 18 / 5 15,195 23.13 s
展开

运行与结束

run key
goal_graph_hybrid__open_evidence-03
termination
model_budget_exhausted
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
0
合并冲突
0

执行指标

重复动作
13
无关动作
0
路径切换
关键路径轮次
5
控制器拒绝
13
prompt / completion
13,973 / 1,222

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · controller P dc987dab7e8e7fd37032a6af207c222031c41edae27db2550ede7bf523b94ecb R 1199fd234b315946d0650fb28abc53628005ff2fa289c4986b1b55c67fc6af9b
  2. round 2 · controller P c1064ea44c0ba9977ba2d6693701e9ab5eb6b16761567b2fc95802f96994a2e1 R 6585026d6b9f78fd25ad14cc61a2819bb6e8c63a13c822b5ea071044fbb9f36c
  3. round 3 · controller P b59dc16772b39e0519277afe51130696af1af154e8cac0eadb8290922d5fdfd5 R 10415679d83912250eaf38bdf50b94ee83b78932b4e178945acdd71f57da1c1e
  4. round 4 · controller P aec937e1d0250652f409f871bc0a302998d4df5d7d08f200e5d7c0305e8817aa R d4d6c7854a49625d9f6289317ef5fd916199cef00e953b7b32b8a65fe956f224
  5. round 5 · controller P 2ca3e64d2ffad3dee107f106a677e868dcc990fad0ff6af3684175e39ba400fa R e4a64fff85f2685a276ebd85d815426187214d6b3a0f4fe482ba7e252e077faf
  6. round 6 · controller P 942638ac3dad7ea0af064265216fe3c64bc6bd0d8a3d7907206c652340312363 R e4a64fff85f2685a276ebd85d815426187214d6b3a0f4fe482ba7e252e077faf
  7. round 7 · controller P 2f24523e0bda95a2dfb67267420e3e6a1838cc2b0c905d90f85edc5e500afeef R e4a64fff85f2685a276ebd85d815426187214d6b3a0f4fe482ba7e252e077faf
  8. round 8 · controller P 8daf0967fab7e901cd4f625ac9fc7682d58e8b929caae833f6b16c0d98ea47f0 R e4a64fff85f2685a276ebd85d815426187214d6b3a0f4fe482ba7e252e077faf
  9. round 9 · controller P 3b2e56f7ee259a548745083d2d8c1bd89ce3795e3ffc90835fb4215272e6bc6b R e4a64fff85f2685a276ebd85d815426187214d6b3a0f4fe482ba7e252e077faf
  10. round 10 · controller P 184573af4131e16be1a4de9695115e15509c07e918e0fee6cc295cea7915dd16 R e4a64fff85f2685a276ebd85d815426187214d6b3a0f4fe482ba7e252e077faf
  11. round 11 · controller P cd206888fee4025ba5421a9db360f4ec5d69d40733dfcd6384c6b29dd4fcebd0 R e4a64fff85f2685a276ebd85d815426187214d6b3a0f4fe482ba7e252e077faf
  12. round 12 · controller P 98b5bcf8134758d6cdedb391413b429b9550f7d00630d8f6f64581d0a7274909 R e4a64fff85f2685a276ebd85d815426187214d6b3a0f4fe482ba7e252e077faf
  13. round 13 · controller P 094b4b241d36ab7b62dc8288bb1aadc5c585851660ea22755c43fbb830a6cd08 R e4a64fff85f2685a276ebd85d815426187214d6b3a0f4fe482ba7e252e077faf
  14. round 14 · controller P 71142a4d0c6c3cd630527ca1d6ea7070166d1c62f237f8672151b6151ba78fd2 R e4a64fff85f2685a276ebd85d815426187214d6b3a0f4fe482ba7e252e077faf
  15. round 15 · controller P 1c2b57c484889a4dd6afb75e7db134fa8d2e102c1d47e619fc5d0919acbbc6ca R e4a64fff85f2685a276ebd85d815426187214d6b3a0f4fe482ba7e252e077faf
  16. round 16 · controller P 936597fb2e83c73b6876773a44ed3f01c58c9bb56dc421ee0f3265577f57d44a R e4a64fff85f2685a276ebd85d815426187214d6b3a0f4fe482ba7e252e077faf
  17. round 17 · controller P 01d72a675b98e57ed07afcf83eb4ba22a17e502a6d4ca095dcd7f5f200b5ece8 R e4a64fff85f2685a276ebd85d815426187214d6b3a0f4fe482ba7e252e077faf
  18. round 18 · controller P a1cb85a4d1f41d5edc2afbc343004ba80e4732ff8eba7ff8284a40138e560c25 R e4a64fff85f2685a276ebd85d815426187214d6b3a0f4fe482ba7e252e077faf
15 开放取证 open_evidence-04 目标图混合执行 通过 100 18 / 5 15,190 24.34 s
展开

运行与结束

run key
goal_graph_hybrid__open_evidence-04
termination
model_budget_exhausted
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
0
合并冲突
0

执行指标

重复动作
13
无关动作
0
路径切换
关键路径轮次
5
控制器拒绝
13
prompt / completion
13,973 / 1,217

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · controller P 92380e127bca408bdf2978ff60b609fd7750dd61cbfee15b4b966ad1d4a2dbfd R 0929f8a7ea7640da08e12673da0f1f33c8572bd3cf1f7721e9a0e0984916ada1
  2. round 2 · controller P 94708bc38ea9b192b05221f1c2fc0b6dbcc8328133ef476858900758fd85e382 R 37ab26215c7e86116bdf3ddf389b2f31ac0384858b248e27ea6656dad134b13b
  3. round 3 · controller P 46839806710ea1308c92c80ea4e3c5d6d76b0abae9e609b6be5ef92ead547165 R 8abfaf49df46d084f807c63cc69f2dacbc004154f756d09d3531f73f38d46083
  4. round 4 · controller P 7f3c897358683016a38bcb6d23c30132ae47bf12b2d9c3cb59b3f086bc23d2f4 R e06aee95a62932cefd87332e72f720254a5ed08ae36d0052d2f24b44cef3a129
  5. round 5 · controller P f512c1900b4254fcde5c3e31f668b2efcc4a9e7eb621b2759867d5fe40bf63c4 R 2fd737c26653eec5b9da0cdae46d9f7d7485e947b31bc64cbef7b515be9425f5
  6. round 6 · controller P 074dd91602715d41a1451e799c07e9e90773d9b03c1c1b28ec0015f9c052666c R 2fd737c26653eec5b9da0cdae46d9f7d7485e947b31bc64cbef7b515be9425f5
  7. round 7 · controller P 305c0631e4568def2b03d4ea08df8f8e7d624358d74ecdd8d15084c32896d1c2 R 2fd737c26653eec5b9da0cdae46d9f7d7485e947b31bc64cbef7b515be9425f5
  8. round 8 · controller P 2c2dad03bce8e17659317629c4b83c4291ee46a29376069baeb3eda0a53327be R 2fd737c26653eec5b9da0cdae46d9f7d7485e947b31bc64cbef7b515be9425f5
  9. round 9 · controller P 0bc7601928152740de6fbacb418918a71f11e9795254fe189cf84c9a742c627b R 2fd737c26653eec5b9da0cdae46d9f7d7485e947b31bc64cbef7b515be9425f5
  10. round 10 · controller P a5c128217689dc08d37e672d348f81dae4640592392973b01bb19ab3364b1554 R 2fd737c26653eec5b9da0cdae46d9f7d7485e947b31bc64cbef7b515be9425f5
  11. round 11 · controller P 9c1638d775fdf8627a97f15910f0f0f84de53fb451601421ef9bf5aaefb60b3f R 2fd737c26653eec5b9da0cdae46d9f7d7485e947b31bc64cbef7b515be9425f5
  12. round 12 · controller P 45342142f558d37174c64879bb086c8b1b7ae92917eeeb367ab0fc3f970d7f57 R 2fd737c26653eec5b9da0cdae46d9f7d7485e947b31bc64cbef7b515be9425f5
  13. round 13 · controller P 7e9c86d869736f04d528d05ac2c08a171b62eb612684afc64dee1908d6eea56d R 2fd737c26653eec5b9da0cdae46d9f7d7485e947b31bc64cbef7b515be9425f5
  14. round 14 · controller P 1000063f35da5560fb16efcfcf73b18a52491f453c8c391c566bfd3dee94879d R 2fd737c26653eec5b9da0cdae46d9f7d7485e947b31bc64cbef7b515be9425f5
  15. round 15 · controller P 827ee78ce5214022761d398a110e724f7ed57557cee3e8981edc98b1b0ea4998 R 2fd737c26653eec5b9da0cdae46d9f7d7485e947b31bc64cbef7b515be9425f5
  16. round 16 · controller P cb8818bfb49782333ebf94181ef4e2ab0fdf64ff46102a41a0af8edb64587577 R 2fd737c26653eec5b9da0cdae46d9f7d7485e947b31bc64cbef7b515be9425f5
  17. round 17 · controller P 7d292b48250b90db7ef27018269e22b34094ab76bd34f669fa135b64987a8cca R 2fd737c26653eec5b9da0cdae46d9f7d7485e947b31bc64cbef7b515be9425f5
  18. round 18 · controller P 29689db41bf5ff361caaae4169099b7be1820e1cf44f9a59a62530a866572e63 R 2fd737c26653eec5b9da0cdae46d9f7d7485e947b31bc64cbef7b515be9425f5
16 开放取证 open_evidence-04 开放式执行 通过 100 6 / 6 4,312 6.34 s
展开

运行与结束

run key
contracted_react__open_evidence-04
termination
finish
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
0
合并冲突
0

执行指标

重复动作
0
无关动作
0
路径切换
关键路径轮次
5
控制器拒绝
0
prompt / completion
4,115 / 197

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · controller P 58f0781c166f3952ea4317417475c39b7f300cf8a72b11b869138c8f8b046e69 R 1199fd234b315946d0650fb28abc53628005ff2fa289c4986b1b55c67fc6af9b
  2. round 2 · controller P 7bfaf958393c788fbfa5c15621ce466895b3f341d905b1204007f611bc3731b1 R 37ab26215c7e86116bdf3ddf389b2f31ac0384858b248e27ea6656dad134b13b
  3. round 3 · controller P f5d7664dbf36b0d1d8e58a33ee9498743c61ed2be7f0d6f73a6449bd9652a257 R 8abfaf49df46d084f807c63cc69f2dacbc004154f756d09d3531f73f38d46083
  4. round 4 · controller P 5c6612d0567e2efe7f8df8991f4461a68e8e4afa6e961f77c6f6451314093517 R e06aee95a62932cefd87332e72f720254a5ed08ae36d0052d2f24b44cef3a129
  5. round 5 · controller P 589b763485107c677a95b0c46b0c9dfc61907b488315d15c2f6f22d8f8314fcd R 2fd737c26653eec5b9da0cdae46d9f7d7485e947b31bc64cbef7b515be9425f5
  6. round 6 · controller P d5ad4f0099d5520dd60a17c8f63a2e9a8b88359254b6cff5dbe9f9844add7f20 R 7310372e895dfcd49057382944f726a4b0ab03cbdea6904b834b3516d3bead52
17 并行拆解 parallel_decomposition-01 开放式执行 未通过 25 6 / 6 4,019 6.46 s
展开

运行与结束

run key
contracted_react__parallel_decomposition-01
termination
finish
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
3
合并冲突
0

执行指标

重复动作
0
无关动作
0
路径切换
关键路径轮次
5
控制器拒绝
0
prompt / completion
3,813 / 206

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · controller P f0bab67495651ae4bcb22f53257d8af4d777f8d78e07a18a48010d99e16bc204 R 1199fd234b315946d0650fb28abc53628005ff2fa289c4986b1b55c67fc6af9b
  2. round 2 · controller P d6e346682f3cb8b594d6de97f135f965a9c17d920cfc4f1425dd385e50331eaa R 1de8849ca69262c0e648cf36bd05d8d9581a3545f073721fad91bb36954de3ca
  3. round 3 · controller P 83fae745b1f9d5dffefdec588a739ec73e5dfb75c64898b184e09ce8a4e55e0a R f13f77122b5d1af690bab4fe8c29a53762f4a826f918eabb67645da48c459cbd
  4. round 4 · controller P e8d9496b3b3b020905f9680d7d1a28aef1579139c914bca569b6e9dc6891a0d0 R f66599ca184ac1930b0d5761d8c44ad03628cb1aef77d215737de5d774ffcdcb
  5. round 5 · controller P 1d59c49d5d81743e26a6f19907b873a4ca23e11f3cb8ba152b0c3dc0ccc25862 R 0e432736d38ed622dfda6f4bd17ee4a286d5a3d19cd34db97a72ba2c45ef0910
  6. round 6 · controller P eb5d48c85da85b0c5d7ea1bbb53eb4db1ede917c393d96a68188106447bc773c R 7310372e895dfcd49057382944f726a4b0ab03cbdea6904b834b3516d3bead52
18 并行拆解 parallel_decomposition-01 目标图混合执行 未通过 25 18 / 5 14,754 20.55 s
展开

运行与结束

run key
goal_graph_hybrid__parallel_decomposition-01
termination
model_budget_exhausted
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
3
合并冲突
0

执行指标

重复动作
13
无关动作
0
路径切换
关键路径轮次
2
控制器拒绝
13
prompt / completion
13,426 / 1,328

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · alpha P abf4be44b6597f6e4306afff3a30c63892ea7717a4a6331fb9431c04c654c245 R 9e7968c4d43b8acfa2488237b84398061841be61d4b7ac91dc30f0047173c0bb
  2. round 1 · beta P f2718f4e414bf87491b96e6de655150318e1bacae142230b29f5ffd2fbfb224d R 14f1f147551bfe49d431a2dfc29768d532d90b992fb959c08cb324b269a6ccb9
  3. round 1 · gamma P b7ac62f2b6d734cbf2518069850031ed88233af3eda64d73254836ad42800b87 R c65f6c95f348f2420eff8c516a08897529fadaec1456a4ca1256f6b1ce59245e
  4. round 2 · controller P afa95fa19f53563fc186b764c8b0d1e7cefb6b9e0074452841f4f70828c37365 R 0e432736d38ed622dfda6f4bd17ee4a286d5a3d19cd34db97a72ba2c45ef0910
  5. round 3 · controller P 54bc6c660a03e54b705304fc8fbe83dca2b8179fed0e5435d010826e35abf15f R 0e432736d38ed622dfda6f4bd17ee4a286d5a3d19cd34db97a72ba2c45ef0910
  6. round 4 · controller P 86bf64e6213639976553f7807656911d176f90b0a1ae523454574d69b777ba1a R 0e432736d38ed622dfda6f4bd17ee4a286d5a3d19cd34db97a72ba2c45ef0910
  7. round 5 · controller P 49a733b2c2ca4ba30bd2beb08191308477637d31c0221836c66c4c665c631243 R 0e432736d38ed622dfda6f4bd17ee4a286d5a3d19cd34db97a72ba2c45ef0910
  8. round 6 · controller P c9fd3a36f4a4b5ca1ab0c27b826e5857c827bd920c4086a3c45eafe2ea4c6c37 R 0e432736d38ed622dfda6f4bd17ee4a286d5a3d19cd34db97a72ba2c45ef0910
  9. round 7 · controller P 73140a882a4f8ee3dd1aec9e827cd6b409b02d55779a3b39a70dbf6fbb5ab5ad R 1199fd234b315946d0650fb28abc53628005ff2fa289c4986b1b55c67fc6af9b
  10. round 8 · controller P 0938e708ba0a84035d4de6c77bd0bc896ff56747f44fb181e49c19b908e05f05 R 0e432736d38ed622dfda6f4bd17ee4a286d5a3d19cd34db97a72ba2c45ef0910
  11. round 9 · controller P 8d49a2d6a8a3c6ca049cffed328742f6ec5ee6c606f8a0f0ca741dfff23a94cf R 0e432736d38ed622dfda6f4bd17ee4a286d5a3d19cd34db97a72ba2c45ef0910
  12. round 10 · controller P cf00744a98027af39ba363fe0ce21fa06b005a9111b10602f1e99f809e985d73 R 0e432736d38ed622dfda6f4bd17ee4a286d5a3d19cd34db97a72ba2c45ef0910
  13. round 11 · controller P fe99616c480af6fee63b85e48f723df37be499d00910bbf61e05893d9c23f8fc R 0e432736d38ed622dfda6f4bd17ee4a286d5a3d19cd34db97a72ba2c45ef0910
  14. round 12 · controller P ccf711ba717f60411d21538ba4e4be2be602f0d2ea74eff8a3d5780b84cc7d33 R 0e432736d38ed622dfda6f4bd17ee4a286d5a3d19cd34db97a72ba2c45ef0910
  15. round 13 · controller P a67d601be45fcb166853d14b55767cc8847b5572233425c1d71ceda57a247663 R 0e432736d38ed622dfda6f4bd17ee4a286d5a3d19cd34db97a72ba2c45ef0910
  16. round 14 · controller P 0683bf2181cbc621357724815b6f200fbb1468ffc9babde39a4a367c4331d24a R 0e432736d38ed622dfda6f4bd17ee4a286d5a3d19cd34db97a72ba2c45ef0910
  17. round 15 · controller P 8b24543e175ff8e7bd610ed3ee2265f71ebd1012ecdf096dc50df156f24c9cbe R 0e432736d38ed622dfda6f4bd17ee4a286d5a3d19cd34db97a72ba2c45ef0910
  18. round 16 · controller P 5829d8da12b9832f96d5338c5704a293e3508491beaf894cf3ac6c534232f005 R 0e432736d38ed622dfda6f4bd17ee4a286d5a3d19cd34db97a72ba2c45ef0910
19 并行拆解 parallel_decomposition-02 目标图混合执行 未通过 25 18 / 5 14,952 20.65 s
展开

运行与结束

run key
goal_graph_hybrid__parallel_decomposition-02
termination
model_budget_exhausted
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
3
合并冲突
0

执行指标

重复动作
13
无关动作
0
路径切换
关键路径轮次
2
控制器拒绝
13
prompt / completion
13,624 / 1,328

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · alpha P b822d1a67e2563f1647bf40d1100964b39c0009529728f11a88080cb94782a01 R 1e36ed3c08e06911f1ee36e88519dcd4141fc0c12adf9b92d9a20852b3b2c66c
  2. round 1 · beta P 3a1f8d6529e6fecb485b38fbb539771ccc7d3cd423399225233fe9c0f2bed88b R 0b8c7284b7dd78bb9746b4817a01f4742be81aa0a20276db7a6b1e98daa4e830
  3. round 1 · gamma P d06206372f5c82d91d29fd87896836a984bcdfe7bd8decd02a6a1aee377bd004 R d0dc6f0fb175bb8dfa45f136879ff89e452c3384aae319d8a252e6892571ae92
  4. round 2 · controller P 7b28c23eac4b4b3d7991984d3a11569444adbdd9bf5429f9096b8a2fbcc571ab R 8f0f630160247947c46f8f4da89cdd26601d9616f93595ea4af72ce84d9f44e2
  5. round 3 · controller P 4ee3d6d62988c3391a27525aeb575fd408affac3d3fc5778b76eaf849cd70822 R 8f0f630160247947c46f8f4da89cdd26601d9616f93595ea4af72ce84d9f44e2
  6. round 4 · controller P 47f8c16577e2c8aea74ca12583cc6174562474c93ef317b8158084391db1a8b6 R 8f0f630160247947c46f8f4da89cdd26601d9616f93595ea4af72ce84d9f44e2
  7. round 5 · controller P 114593a7e8efc7913754ca1c0a365597dfd5284b3c40a9887224e0efad6a7cfa R 8f0f630160247947c46f8f4da89cdd26601d9616f93595ea4af72ce84d9f44e2
  8. round 6 · controller P a995320c4ce999c54a849e1563871719bf109a5e2051c3c1d93dbe2e7967a9d1 R 8f0f630160247947c46f8f4da89cdd26601d9616f93595ea4af72ce84d9f44e2
  9. round 7 · controller P fda32afaf1197155720ca08df90d0d27fbc13494e718efaa9f3ecbd8d246bcc4 R 8f0f630160247947c46f8f4da89cdd26601d9616f93595ea4af72ce84d9f44e2
  10. round 8 · controller P 1889bd483bec4cbb2096ab4acdfe6c9cbb261283b9cf70ab42f362ebde5a6e2c R 8f0f630160247947c46f8f4da89cdd26601d9616f93595ea4af72ce84d9f44e2
  11. round 9 · controller P 26fadfb752644b166dc0772aa3cec652355db94fc84d2471b5f0491693b4c020 R 8f0f630160247947c46f8f4da89cdd26601d9616f93595ea4af72ce84d9f44e2
  12. round 10 · controller P 9f76285a64ee33512b9a9c8b250d0a75b9a7d3eaac174af4eace7a3ba79ea761 R 8f0f630160247947c46f8f4da89cdd26601d9616f93595ea4af72ce84d9f44e2
  13. round 11 · controller P a65961b161548c4ce1ee75bb76b6c99c7e7a12d4ad234dc7afc3854eaaa69af0 R 8f0f630160247947c46f8f4da89cdd26601d9616f93595ea4af72ce84d9f44e2
  14. round 12 · controller P c2c034ee4bbb89a14401de3c7cf659270f4fae82cfba6684ac0358ef487db8a3 R 8f0f630160247947c46f8f4da89cdd26601d9616f93595ea4af72ce84d9f44e2
  15. round 13 · controller P 3ddf474bfdac8a71656448cb1920f3084db6d1b8d4f1ec986c8973bfa4d03b36 R 1199fd234b315946d0650fb28abc53628005ff2fa289c4986b1b55c67fc6af9b
  16. round 14 · controller P 5839a5f2b7808c97f67c0d096c16459070984856d7de1d0f0017d8622aed1902 R 8f0f630160247947c46f8f4da89cdd26601d9616f93595ea4af72ce84d9f44e2
  17. round 15 · controller P 1b5f5a295275edfcc4e846bd85febd120d0ec5e538b64ba9f501a252507cba53 R 8f0f630160247947c46f8f4da89cdd26601d9616f93595ea4af72ce84d9f44e2
  18. round 16 · controller P b5054cec787dc2c3b3b0eded39a19422f1b290c637372dfb60717793f98b85f1 R 8f0f630160247947c46f8f4da89cdd26601d9616f93595ea4af72ce84d9f44e2
20 并行拆解 parallel_decomposition-02 开放式执行 未通过 25 6 / 6 4,019 5.74 s
展开

运行与结束

run key
contracted_react__parallel_decomposition-02
termination
finish
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
3
合并冲突
0

执行指标

重复动作
0
无关动作
0
路径切换
关键路径轮次
5
控制器拒绝
0
prompt / completion
3,813 / 206

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · controller P 942b9be2dbe2660fda6f024cb8d8549e820367db33e9434b61bfd2960d3d804c R 1199fd234b315946d0650fb28abc53628005ff2fa289c4986b1b55c67fc6af9b
  2. round 2 · controller P 9388bddf5c1d0b2ccce2383a413ff7f82834fe575f9d48779e963b1dafe78f26 R 447dbd199d303d82d66e9427d5d0fe13dc522a29a9ea4403757d0ced5ff550f8
  3. round 3 · controller P 2a5f2798cf53441b1212d103e20d9fc6371b4406a226ccd0c078ffd34b850709 R 715f2a7f205bb8ddf07587561c857e485ab369112132673995d233e265b93c11
  4. round 4 · controller P ac4e78c84b0d9930e288532d2dbd57ad0b20f61d29327d03662aa343bf4db64a R f5fd750b6de3203b71c7ad76ab3660e1a4ad89b61767d6bdc3ba7db89d71ba0b
  5. round 5 · controller P ce570a1d01227dfc8a3304e74ba53c9a04cd778f1adf66e8e8fd34e919b20e1b R 8f0f630160247947c46f8f4da89cdd26601d9616f93595ea4af72ce84d9f44e2
  6. round 6 · controller P 740f6c116ee5d780c549ccaf644cdb136edbef05831bf3a996c419d606d24c90 R 7310372e895dfcd49057382944f726a4b0ab03cbdea6904b834b3516d3bead52
21 并行拆解 parallel_decomposition-03 开放式执行 未通过 25 6 / 6 4,019 5.46 s
展开

运行与结束

run key
contracted_react__parallel_decomposition-03
termination
finish
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
3
合并冲突
0

执行指标

重复动作
0
无关动作
0
路径切换
关键路径轮次
5
控制器拒绝
0
prompt / completion
3,813 / 206

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · controller P b8d833d5b85debe6321702252944c2d9e71854887a7ad81b3a685a9d5c3a3d3f R 1199fd234b315946d0650fb28abc53628005ff2fa289c4986b1b55c67fc6af9b
  2. round 2 · controller P c6147add13e60e396a3327ee0ea6898776b20c7c3c11f86a4f6e471093dabcd5 R df344990177ddf50d3cabf583e8d104941fb557dc066b642449c918a2f82095a
  3. round 3 · controller P 427d3c75d21c745b7da334944d512072ff5e5a8969cf805a45929488a80c12ce R 8aaf01ae9b648ee883a6281468ae3b7c9ad597bf0cbf963bf441c34c94098281
  4. round 4 · controller P d7b6d5b77e671b2aac5c24d9b45b3c2092f47b65af4d4e132bcd63d8044fec89 R 992e9b146d52d41e318ff3232285bd8c0576399a8a0c4a5fc811f1ebd06e6ca9
  5. round 5 · controller P 34b9f19b0dc75f553a87bca30e8249a3368207051a2a6353498afa446e54ce18 R e75bca37b0a9c98cfde29535bee930bb64787e7a3ce0feb977f1c175c794ee92
  6. round 6 · controller P 428d20265e2570f345ee7691bd4471497b2698eb797900efa1c63c9f46cea3e7 R 7310372e895dfcd49057382944f726a4b0ab03cbdea6904b834b3516d3bead52
22 并行拆解 parallel_decomposition-03 目标图混合执行 未通过 25 18 / 5 14,721 18.59 s
展开

运行与结束

run key
goal_graph_hybrid__parallel_decomposition-03
termination
model_budget_exhausted
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
3
合并冲突
0

执行指标

重复动作
13
无关动作
0
路径切换
关键路径轮次
2
控制器拒绝
13
prompt / completion
13,393 / 1,328

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · alpha P 3111bd0ba295dfc91daf75b48c1d74b8d3c22a23d110b86731e45ff9ccd2ca1c R 0df103790891a9fd1570a3ec98d04c624411e34e235100c42406abf4a89a72b1
  2. round 1 · beta P 2aae6bb4ddce830791cc218c96ac59337725177ae8bb36d9c4745e124851ca35 R ec759c082dfbd0137a5422edd12893495649a5dcaf3fadc143692db79f7f110c
  3. round 1 · gamma P 07c8d3580822d7d179e0d9262b84a496fce23534ed9b87e1270dd122a15ce074 R 82449bc20814c0e570f209c347001bae7d334adffaec51680acdfcdd2d317c4a
  4. round 2 · controller P 192b3e9814a23d92d2e5d0ca8a743809cae64c41a3608c4662c66610b80f5c7e R e75bca37b0a9c98cfde29535bee930bb64787e7a3ce0feb977f1c175c794ee92
  5. round 3 · controller P f064212556167dee024fdf73c5bd734e50bdbb59bea201f0dc115348054a67f9 R e75bca37b0a9c98cfde29535bee930bb64787e7a3ce0feb977f1c175c794ee92
  6. round 4 · controller P 1097c86f6714077dade9b045fcddda239f43c42c0ac1ff4531ed6afd8a192b78 R e75bca37b0a9c98cfde29535bee930bb64787e7a3ce0feb977f1c175c794ee92
  7. round 5 · controller P 83c93d781458c4d6ddac681fcb34d32cc1d283c4efdcf42b92fc85b459924241 R e75bca37b0a9c98cfde29535bee930bb64787e7a3ce0feb977f1c175c794ee92
  8. round 6 · controller P 207379d3e1698913379647eb82b069e35d343454988fcd3ed0a65c4aa5fa3b72 R 1199fd234b315946d0650fb28abc53628005ff2fa289c4986b1b55c67fc6af9b
  9. round 7 · controller P 29a7251c475001fc981d1f2d32daf23fecfa16f1da09e0f7cce429b35c2f0c26 R e75bca37b0a9c98cfde29535bee930bb64787e7a3ce0feb977f1c175c794ee92
  10. round 8 · controller P 8f6ba21f41fee9a193b6ef38ebf4395855d18559d98839e1bee1ea4a0f78baaa R e75bca37b0a9c98cfde29535bee930bb64787e7a3ce0feb977f1c175c794ee92
  11. round 9 · controller P c6367b436090f1582fb0cffaf69b2e25819270719a7c8702347a801c24e598f5 R e75bca37b0a9c98cfde29535bee930bb64787e7a3ce0feb977f1c175c794ee92
  12. round 10 · controller P 72d8c7ab1ae74b50f2de075ca484494bb48c2e490895a18499e6b89a5bc259d8 R e75bca37b0a9c98cfde29535bee930bb64787e7a3ce0feb977f1c175c794ee92
  13. round 11 · controller P 1b70b898cd21893bc20f154010d79aa689f3f1ab11aee9e059d72ac6b6d0065d R e75bca37b0a9c98cfde29535bee930bb64787e7a3ce0feb977f1c175c794ee92
  14. round 12 · controller P da0d0cf4200fc01c39d42736b1b963ccbd0c87dbc34e29631cd0f6391ea0d34e R e75bca37b0a9c98cfde29535bee930bb64787e7a3ce0feb977f1c175c794ee92
  15. round 13 · controller P cbc2e99f3569ac6696a0e5b309641ef1dbde5d9e555beb93b1dcc9f86ee4f091 R e75bca37b0a9c98cfde29535bee930bb64787e7a3ce0feb977f1c175c794ee92
  16. round 14 · controller P 52b3b7fb3fb15ca3b3201b7a3e88d722aad13fc5dce125a0ed104c9a119708a7 R e75bca37b0a9c98cfde29535bee930bb64787e7a3ce0feb977f1c175c794ee92
  17. round 15 · controller P da79cb8156f355e569fe60e79e2a795542fb507d6e01aa18a263b0f4b691028c R e75bca37b0a9c98cfde29535bee930bb64787e7a3ce0feb977f1c175c794ee92
  18. round 16 · controller P fe713bf0e1a8a270085650b8f0d21c79e26576dcf8c4949728b752aa703cd5b6 R e75bca37b0a9c98cfde29535bee930bb64787e7a3ce0feb977f1c175c794ee92
23 并行拆解 parallel_decomposition-04 目标图混合执行 未通过 25 18 / 4 15,136 19.17 s
展开

运行与结束

run key
goal_graph_hybrid__parallel_decomposition-04
termination
model_budget_exhausted
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
3
合并冲突
0

执行指标

重复动作
14
无关动作
0
路径切换
关键路径轮次
2
控制器拒绝
14
prompt / completion
13,723 / 1,413

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · alpha P 22cefbeacdb4e042c69a4a2cc18a646ee9f9e8e075dfb4eabe04d1a1161622a0 R f069d18237a002163bab29c8f4fe01f3128fb22845f41ed26f3c19d57c571855
  2. round 1 · beta P cb68447e3d6fbc6032219ed40a9870d14d62a37c19d61861a6da26e6b08921d9 R fc68ccd8ca6ec71ee38d69393c16e1eb330aa082acdd91ebec2343956e2caab8
  3. round 1 · gamma P 430be591625741f180efd6ebde5742d3b895d75689aaa6a04c37059f80ed2845 R d27b118d9548b9f99df00ebf78eb55381c6606e4fe0e65d17d174b623544d1d3
  4. round 2 · controller P 0e897cd5539c3646e384f8ca939312cb8af63dff2f13fb2f73ba9b48d240fd36 R c3167d0a7630e501e407451749c68a74c38200d605a4ea5c50fd3c7db6fa489e
  5. round 3 · controller P cccacb74a482cbb0f342a28e1af1196bc5c1d3a101c4ac3ef02b059176bb8a8d R c3167d0a7630e501e407451749c68a74c38200d605a4ea5c50fd3c7db6fa489e
  6. round 4 · controller P 14229f2716a7b5c1531fae8d393ff1d9979149306c7d79ebeeb7da1fb496a552 R c3167d0a7630e501e407451749c68a74c38200d605a4ea5c50fd3c7db6fa489e
  7. round 5 · controller P d996a6ee110075cc494d9b6a971ce3e87344a15ff3b40987bd43ccca609dfbaa R c3167d0a7630e501e407451749c68a74c38200d605a4ea5c50fd3c7db6fa489e
  8. round 6 · controller P e7a5f9f920ec800db7a22d34b8e64fe3ede13c2c54b748c507c58f638d54228b R c3167d0a7630e501e407451749c68a74c38200d605a4ea5c50fd3c7db6fa489e
  9. round 7 · controller P 8770774811bbf9d2190fc9afc27004c94fbb2b9e766132439005382ba56c4c4f R c3167d0a7630e501e407451749c68a74c38200d605a4ea5c50fd3c7db6fa489e
  10. round 8 · controller P 7f8f10b6ce8be48d2fc8609e1e86a854dc7504b61de174b62e4d859b29aea06a R c3167d0a7630e501e407451749c68a74c38200d605a4ea5c50fd3c7db6fa489e
  11. round 9 · controller P d13c52a2e9ed8d31d418c481b5391892e7e0e930a1b1d0d16e3ed7d420461377 R c3167d0a7630e501e407451749c68a74c38200d605a4ea5c50fd3c7db6fa489e
  12. round 10 · controller P ab99438feeb323b287b8bbc5bce0c1ad2dd37d047abf6f31ea7e1f13acd8b613 R c3167d0a7630e501e407451749c68a74c38200d605a4ea5c50fd3c7db6fa489e
  13. round 11 · controller P 665df71870af37eeed9241c4e16f6b04e0cdd1ae855277f1117a3e95111b3894 R c3167d0a7630e501e407451749c68a74c38200d605a4ea5c50fd3c7db6fa489e
  14. round 12 · controller P d123f4a658faaac1a144673ce61583aad823ea6e38e772dd7d6bdbdc119a4a6f R c3167d0a7630e501e407451749c68a74c38200d605a4ea5c50fd3c7db6fa489e
  15. round 13 · controller P df5a9c5473031d36696c83833518c81e2820f7f042761a142e63060a0f8bf60c R c3167d0a7630e501e407451749c68a74c38200d605a4ea5c50fd3c7db6fa489e
  16. round 14 · controller P 6e3c3f5b824f946af7244ae1f067334eb7642090c74e9fea94ac5b85cb598022 R c3167d0a7630e501e407451749c68a74c38200d605a4ea5c50fd3c7db6fa489e
  17. round 15 · controller P df7e383685983fa1e7d33b4bc96c958a257e1b62d5c0738c4a2c2e6c8d60b146 R c3167d0a7630e501e407451749c68a74c38200d605a4ea5c50fd3c7db6fa489e
  18. round 16 · controller P 58f010e73f172707b7e599d248bcf57c88be5077012c27889001ffa6640e5af7 R c3167d0a7630e501e407451749c68a74c38200d605a4ea5c50fd3c7db6fa489e
24 并行拆解 parallel_decomposition-04 开放式执行 未通过 25 6 / 6 4,019 7.14 s
展开

运行与结束

run key
contracted_react__parallel_decomposition-04
termination
finish
eligibility
eligible_agent_result
pair
eligible_pair

代码验收

关键事实覆盖
100 %
错误证据
0
合并遗漏
3
合并冲突
0

执行指标

重复动作
0
无关动作
0
路径切换
关键路径轮次
5
控制器拒绝
0
prompt / completion
3,813 / 206

模型交换哈希

用于核对公开记录,不公开响应正文。

  1. round 1 · controller P 7c5c5e997fa7ff726801a0d1577ac4e443c9f44cc6a67298d5fff0ba4cd6d1de R 1199fd234b315946d0650fb28abc53628005ff2fa289c4986b1b55c67fc6af9b
  2. round 2 · controller P 6e853cc1c1f87ba223bbdc5c589eefc0b7db695f53491d5e829b6dba3a2517e8 R f069d18237a002163bab29c8f4fe01f3128fb22845f41ed26f3c19d57c571855
  3. round 3 · controller P feabca689aa94bf141752cb2e1afb442401c8bc3c572f159216a8bdf1c107c4d R 253e7aefd335d10a3c2d4e3d15717bfc9b8cfe0b1a2ad7c53e4bb54c03beb3fb
  4. round 4 · controller P bc64e54bc1a6369514298f8741faafef4545acd63a5cd01378a0fe8767d60f34 R e22918c728ab35f3298e311aadf9c35946e5116744300e481fd419ee40887ca9
  5. round 5 · controller P bf745d54999f6371ec6614d139f49a5caffca12f5d696ac88d627e6250a26549 R c3167d0a7630e501e407451749c68a74c38200d605a4ea5c50fd3c7db6fa489e
  6. round 6 · controller P 53f6eadb7759b485991aeafb370ba795dc86e1d2c2979b3aaae4aa191e693418 R 7310372e895dfcd49057382944f726a4b0ab03cbdea6904b834b3516d3bead52

03 · Download

下载完整公开证据包

manifest 是完整性边界:它记录其余五个文件的字节数和 SHA-256。下载后可自行复算。

批次2026-07-22-pilot-b-deepseek-v4-flash-a

模型身份deepseek-v4-flash

公开规则字段白名单投影;不含私有输入、原始响应正文和凭据。

查看实验代码与复现命令