ENT-b2bceffe-2ce0-475c-b161-529c13553b46 · benchmark · software issue resolution benchmark subset
SWE-bench Verified
Reviewed identifier SWE-bench_Verified · current status intentionally unknown
Scope
Reviewed benchmark or evaluation-protocol identity; model results and leaderboard ranks are separate, unrepresented observations.
Boundary
Verified means the subset was reviewed for solvability; it is not verification of a submitted model result, software quality or production engineering ability.
Benchmark comparability passport
- Publisher
- SWE-bench
- Version
- Verified subset; source revision 02e7a74f
- Task scope
- Subset of 500 software-repository problems confirmed by software engineers as solvable.
- Metric contract
- Per-instance resolved/not-resolved outcome produced by the selected SWE-bench evaluation harness; no cross-harness aggregate is represented.
- Evaluation conditions
- Result identity requires exact dataset revision, task repository state, prediction patch, container image/harness version, run ID and grading logs.
- Comparability boundary
- Do not compare results across dataset, task-repository, image, harness or grading revisions without an explicit bridge; cached run IDs can reuse earlier results even when predictions change.
- Contamination boundary
- The selected source does not establish absence of issue, patch or repository exposure in model training or tool retrieval.
This passport describes a benchmark definition or protocol. It publishes no model score, rank or cross-version equivalence.
Source-qualified relations
No reviewed relation is asserted for this identity.
Atlas claims with an exact source route
No exact source-URL connection in r0056. No name match was substituted.
Every displayed attribute links to source-qualified statement IDs. Relations are explicit and source-qualified. Identity does not imply ownership, legal status, present availability or equivalence with a mutable alias.