{"abstract":"A late gradient from an older collective step is applied.","category":"Distributed training aggregation","checks":6,"contract":"Given [step,worker,gradient] contributions and target step, return sum for exactly that step.","contract_signature":"rows, step","evaluation_group":"w3-dta-stale-step-filter","failed_approach":"Choosing the maximum observed step discards valid current-step workers when one future item arrives.","family":"w3-distributed-training-aggregation-stale-step-filter","id":"FA-99281","implementations":{"attempt":{"sha256":"d8ef020d45c0e49cf3ff5a7c6d45e88e7f19363baccf88f43859aea462599d35","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(rows, step):\n    newest=max((s for s,w,g in rows),default=step); return sum(g for s,w,g in rows if s==newest)\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\nrows0=[[N,'a',N],[N,'b',2*N],[N-1,'late',100*N]]\ncheck('old step excluded',solve(rows0,N),3*N)\ncheck('empty',solve([],N),0)\ncheck('no target entries',solve([[N+1,'x',N]],N),0)\ncheck('one current entry',solve([[N,'x',4]],N),4)\ncheck('future does not replace target',solve([[N,'a',N],[N+1,'b',20*N]],N),N)\ncheck('negative contributions',solve([[N,'a',-N],[N,'b',2*N]],N),N)\nprint(json.dumps({\"observations\": observations, \"passed\": all(x[\"passed\"] for x in observations)}, ensure_ascii=False))\nraise SystemExit(0 if all(x[\"passed\"] for x in observations) else 1)\n"},"broken":{"sha256":"568b3cc54f4dfadc1718771687c0cd5a954cfe7941c1a4440c44783e29827311","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(rows, step):\n    return sum(g for s,w,g in rows)\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\nrows0=[[N,'a',N],[N,'b',2*N],[N-1,'late',100*N]]\ncheck('old step excluded',solve(rows0,N),3*N)\ncheck('empty',solve([],N),0)\ncheck('no target entries',solve([[N+1,'x',N]],N),0)\ncheck('one current entry',solve([[N,'x',4]],N),4)\ncheck('future does not replace target',solve([[N,'a',N],[N+1,'b',20*N]],N),N)\ncheck('negative contributions',solve([[N,'a',-N],[N,'b',2*N]],N),N)\nprint(json.dumps({\"observations\": observations, \"passed\": all(x[\"passed\"] for x in observations)}, ensure_ascii=False))\nraise SystemExit(0 if all(x[\"passed\"] for x in observations) else 1)\n"}},"limitations":"Inputs are already materialized Python values from a fixed toy collective; no transport, failure recovery, optimizer state, or real model behavior is represented. This reproducer isolates one failure mechanism. Results cover the supplied fixtures. Variants within a family share a test contract and should remain grouped when constructing evaluation splits. Related mechanisms with a shared evaluation_group must also remain together; these controlled models are not independent production incidents.","method":"Deterministic executable model with adversarial boundary fixtures.","provenance":{"created_by":"Failure Map","dependencies":"Python standard library","family":"w3-distributed-training-aggregation-stale-step-filter","generated_at":"2026-09-30T16:48:22.633661+00:00","license":"CC0-1.0","python":"3.12.14","seed":1,"split":"open-access"},"relevance":"A toy distributed-training contract makes replica and shard contribution rules explicit without claiming framework compatibility or performance.","root_cause":"Buffered contributions can arrive after a newer step has started.","sha256":"ee18f317dcdbf8a8082668a55fee1a9803f39b15a4bf627438d579a856789c97","title":"A late gradient from an older collective step is applied · case 01","variant":1,"variant_policy":"The fixture code references N, supplied as 1 through 5. Fixed boundary checks may repeat. These records share one mechanism and must remain together in evaluation splits.","verified":true,"visibility":"public","verification":{"attempt":{"elapsed_ms":29.602,"exit_code":1,"observations":[{"actual":3,"check":"old step excluded","expected":3,"passed":true},{"actual":0,"check":"empty","expected":0,"passed":true},{"actual":1,"check":"no target entries","expected":0,"passed":false},{"actual":4,"check":"one current entry","expected":4,"passed":true},{"actual":20,"check":"future does not replace target","expected":1,"passed":false},{"actual":1,"check":"negative contributions","expected":1,"passed":true}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"old step excluded\", \"actual\": 3, \"expected\": 3, \"passed\": true}, {\"check\": \"empty\", \"actual\": 0, \"expected\": 0, \"passed\": true}, {\"check\": \"no target entries\", \"actual\": 1, \"expected\": 0, \"passed\": false}, {\"check\": \"one current entry\", \"actual\": 4, \"expected\": 4, \"passed\": true}, {\"check\": \"future does not replace target\", \"actual\": 20, \"expected\": 1, \"passed\": false}, {\"check\": \"negative contributions\", \"actual\": 1, \"expected\": 1, \"passed\": true}], \"passed\": false}\n"},"broken":{"elapsed_ms":28.291,"exit_code":1,"observations":[{"actual":103,"check":"old step excluded","expected":3,"passed":false},{"actual":0,"check":"empty","expected":0,"passed":true},{"actual":1,"check":"no target entries","expected":0,"passed":false},{"actual":4,"check":"one current entry","expected":4,"passed":true},{"actual":21,"check":"future does not replace target","expected":1,"passed":false},{"actual":1,"check":"negative contributions","expected":1,"passed":true}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"old step excluded\", \"actual\": 103, \"expected\": 3, \"passed\": false}, {\"check\": \"empty\", \"actual\": 0, \"expected\": 0, \"passed\": true}, {\"check\": \"no target entries\", \"actual\": 1, \"expected\": 0, \"passed\": false}, {\"check\": \"one current entry\", \"actual\": 4, \"expected\": 4, \"passed\": true}, {\"check\": \"future does not replace target\", \"actual\": 21, \"expected\": 1, \"passed\": false}, {\"check\": \"negative contributions\", \"actual\": 1, \"expected\": 1, \"passed\": true}], \"passed\": false}\n"}},"member_only":{"stages":["fixed"],"fields":["implementations.fixed","verification.fixed","harness","repair"],"note":"The verified repair, its recorded checks, the repair description, and the scoring harness are available to members."}}