iiRecord
Agentic AI Atlas · SWE-bench original test set
test-set:swe-bench-originala5c.ai
II.
TestSet JSON

test-set:swe-bench-original

Structured · live

SWE-bench original test set json

Inspect the normalized record payload exactly as the atlas UI reads it.

File · benchmarks/test-sets/source-backed-benchmark-test-sets.yamlCluster · benchmarks
Record JSON
{
  "id": "test-set:swe-bench-original",
  "_kind": "TestSet",
  "_file": "benchmarks/test-sets/source-backed-benchmark-test-sets.yaml",
  "_cluster": "benchmarks",
  "attributes": {
    "displayName": "SWE-bench original test set",
    "benchmarkId": "benchmark:swe-bench",
    "caseCount": 2294,
    "releasedAt": "2023-10-10",
    "composition": "Original SWE-bench issue/PR task collection from 12 popular Python\nrepositories. Each task asks an agent to patch a real repository at\na base commit so fail-to-pass tests succeed.\n",
    "homepageUrl": "https://www.swebench.com/original.html",
    "description": "Source-backed canonical SWE-bench artifact for full real-world\nGitHub issue resolution evaluation.\n"
  },
  "outgoingEdges": [
    {
      "from": "test-set:swe-bench-original",
      "to": "benchmark:swe-bench",
      "kind": "belongs_to_benchmark"
    }
  ],
  "incomingEdges": []
}