Hugging Face's logo

Spaces:
Yashp2003
/
repro-swe-bench-pro-can-ai-agents-solve-long-horizon-software-engineering-tasks
Running

App Files Community
Yashp2003 commited on
Commit
fbc261c
·
verified ·
1 Parent(s): ef3f9a0

Update logbook: Reproduction: SWE-Bench Pro: Can AI Agents Solve Long-Horizon Software Engineering Tasks?

Browse files
Files changed (3) hide show
  1. logbook.json +9 -3
  2. pages/deep-verification-contamination-resistance-agent-solve-rate-and-eval-repo-evidence/page.md +1 -0
  3. pages/index.md +1 -0
logbook.json CHANGED
@@ -8,7 +8,7 @@
8
  "icml2026-repro",
9
  "paper-uEVTdoAbnK"
10
  ],
11
- "updated_at": "2026-07-19T12:06:32+00:00",
12
  "root": {
13
  "slug": "index",
14
  "title": "Reproduction: SWE-Bench Pro: Can AI Agents Solve Long-Horizon Software Engineering Tasks?",
@@ -55,9 +55,15 @@
55
  "title": "Conclusion",
56
  "file": "pages/conclusion/page.md",
57
  "children": []
 
 
 
 
 
 
58
  }
59
  ]
60
  },
61
- "agent_view_tokens": 9087,
62
- "revision": "1784462792555886097"
63
  }
 
8
  "icml2026-repro",
9
  "paper-uEVTdoAbnK"
10
  ],
11
+ "updated_at": "2026-07-19T13:01:37+00:00",
12
  "root": {
13
  "slug": "index",
14
  "title": "Reproduction: SWE-Bench Pro: Can AI Agents Solve Long-Horizon Software Engineering Tasks?",
 
55
  "title": "Conclusion",
56
  "file": "pages/conclusion/page.md",
57
  "children": []
58
+ },
59
+ {
60
+ "slug": "deep-verification-contamination-resistance-agent-solve-rate-and-eval-repo-evidence",
61
+ "title": "Deep Verification: Contamination Resistance, Agent Solve Rate, and Eval Repo Evidence",
62
+ "file": "pages/deep-verification-contamination-resistance-agent-solve-rate-and-eval-repo-evidence/page.md",
63
+ "children": []
64
  }
65
  ]
66
  },
67
+ "agent_view_tokens": 9198,
68
+ "revision": "1784466097528170082"
69
  }
pages/deep-verification-contamination-resistance-agent-solve-rate-and-eval-repo-evidence/page.md ADDED
@@ -0,0 +1 @@
 
 
1
+ # Deep Verification: Contamination Resistance, Agent Solve Rate, and Eval Repo Evidence
pages/index.md CHANGED
@@ -13,3 +13,4 @@
13
  | [Claim 4: All SWE-Bench Pro tasks underwent human verification for adequate resolution context (abstract only).](#/claim-4-all-swe-bench-pro-tasks-underwent-human-verification-for-adequate-resolution-context-abstract-only) |
14
  | [Claim 5: SWE-Bench Pro is a contamination-resistant testbed spanning business applications, B2B services, and developer tools (abstract only).](#/claim-5-swe-bench-pro-is-a-contamination-resistant-testbed-spanning-business-applications-b2b-services-and-developer-tools-abstract-only) |
15
  | [Conclusion](#/conclusion) |
 
 
13
  | [Claim 4: All SWE-Bench Pro tasks underwent human verification for adequate resolution context (abstract only).](#/claim-4-all-swe-bench-pro-tasks-underwent-human-verification-for-adequate-resolution-context-abstract-only) |
14
  | [Claim 5: SWE-Bench Pro is a contamination-resistant testbed spanning business applications, B2B services, and developer tools (abstract only).](#/claim-5-swe-bench-pro-is-a-contamination-resistant-testbed-spanning-business-applications-b2b-services-and-developer-tools-abstract-only) |
15
  | [Conclusion](#/conclusion) |
16
+ | [Deep Verification: Contamination Resistance, Agent Solve Rate, and Eval Repo Evidence](#/deep-verification-contamination-resistance-agent-solve-rate-and-eval-repo-evidence) |