{"abstract":"A flushed partial window is divided by configured capacity.","category":"Gradient accumulation weighting","checks":8,"contract":"gradient_sum is the sum for examples seen in a window; examples is the actual count and capacity is only the maximum. Normalize by actual examples.","contract_signature":"gradient_sum, examples, capacity","evaluation_group":"gradient-accumulation-partial_window","failed_approach":"Capping the count at capacity under-normalizes when a window has accumulated more examples than its nominal threshold.","family":"w3-gradient-accumulation-weighting-partial-window","id":"FA-98861","implementations":{"attempt":{"sha256":"ee6a4ee5679030524aa4135999322f469ab52a30e45525a22b1af1929e614084","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(gradient_sum, examples, capacity):\n    denominator=max(1,min(examples,capacity))\n    return gradient_sum / denominator\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\ncheck('one of capacity three',solve(6*N,N,3),6)\ncheck('full window',solve(12*N,2*N,2),6)\ncheck('empty flush',solve(0,0,4),0)\ncheck('capacity one',solve(7*N,N,1),7)\ncheck('two examples partial',solve(8*N,2*N,5),4)\ncheck('single capacity',solve(9*N,3*N,1),3)\ncheck('zero sum',solve(0,3*N,4),0)\ncheck('count after threshold',solve(12*N,4*N,3),3)\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":"d4ee4ff2778d926619cb0a37c049a788c9f59d270be169f2acc900e818bf619b","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(gradient_sum, examples, capacity):\n    return gradient_sum / max(1,capacity)\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\ncheck('one of capacity three',solve(6*N,N,3),6)\ncheck('full window',solve(12*N,2*N,2),6)\ncheck('empty flush',solve(0,0,4),0)\ncheck('capacity one',solve(7*N,N,1),7)\ncheck('two examples partial',solve(8*N,2*N,5),4)\ncheck('single capacity',solve(9*N,3*N,1),3)\ncheck('zero sum',solve(0,3*N,4),0)\ncheck('count after threshold',solve(12*N,4*N,3),3)\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 scalar or short-list summaries and do not model autodiff, devices, or an optimizer. 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-gradient-accumulation-weighting-partial-window","generated_at":"2026-09-30T16:48:17.605270+00:00","license":"CC0-1.0","python":"3.12.14","seed":1,"split":"open-access"},"relevance":"A deterministic toy model isolates one local accumulation contract.","root_cause":"The configured capacity is used as if every flushed window were full.","sha256":"24bee2db1a44e9897548e6f864f96b0fd85aa2993f95e9a074ba50ffc5aa486c","title":"A flushed partial window is divided by configured capacity · 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":26.18,"exit_code":1,"observations":[{"actual":6.0,"check":"one of capacity three","expected":6,"passed":true},{"actual":6.0,"check":"full window","expected":6,"passed":true},{"actual":0.0,"check":"empty flush","expected":0,"passed":true},{"actual":7.0,"check":"capacity one","expected":7,"passed":true},{"actual":4.0,"check":"two examples partial","expected":4,"passed":true},{"actual":9.0,"check":"single capacity","expected":3,"passed":false},{"actual":0.0,"check":"zero sum","expected":0,"passed":true},{"actual":4.0,"check":"count after threshold","expected":3,"passed":false}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"one of capacity three\", \"actual\": 6.0, \"expected\": 6, \"passed\": true}, {\"check\": \"full window\", \"actual\": 6.0, \"expected\": 6, \"passed\": true}, {\"check\": \"empty flush\", \"actual\": 0.0, \"expected\": 0, \"passed\": true}, {\"check\": \"capacity one\", \"actual\": 7.0, \"expected\": 7, \"passed\": true}, {\"check\": \"two examples partial\", \"actual\": 4.0, \"expected\": 4, \"passed\": true}, {\"check\": \"single capacity\", \"actual\": 9.0, \"expected\": 3, \"passed\": false}, {\"check\": \"zero sum\", \"actual\": 0.0, \"expected\": 0, \"passed\": true}, {\"check\": \"count after threshold\", \"actual\": 4.0, \"expected\": 3, \"passed\": false}], \"passed\": false}\n"},"broken":{"elapsed_ms":28.265,"exit_code":1,"observations":[{"actual":2.0,"check":"one of capacity three","expected":6,"passed":false},{"actual":6.0,"check":"full window","expected":6,"passed":true},{"actual":0.0,"check":"empty flush","expected":0,"passed":true},{"actual":7.0,"check":"capacity one","expected":7,"passed":true},{"actual":1.6,"check":"two examples partial","expected":4,"passed":false},{"actual":9.0,"check":"single capacity","expected":3,"passed":false},{"actual":0.0,"check":"zero sum","expected":0,"passed":true},{"actual":4.0,"check":"count after threshold","expected":3,"passed":false}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"one of capacity three\", \"actual\": 2.0, \"expected\": 6, \"passed\": false}, {\"check\": \"full window\", \"actual\": 6.0, \"expected\": 6, \"passed\": true}, {\"check\": \"empty flush\", \"actual\": 0.0, \"expected\": 0, \"passed\": true}, {\"check\": \"capacity one\", \"actual\": 7.0, \"expected\": 7, \"passed\": true}, {\"check\": \"two examples partial\", \"actual\": 1.6, \"expected\": 4, \"passed\": false}, {\"check\": \"single capacity\", \"actual\": 9.0, \"expected\": 3, \"passed\": false}, {\"check\": \"zero sum\", \"actual\": 0.0, \"expected\": 0, \"passed\": true}, {\"check\": \"count after threshold\", \"actual\": 4.0, \"expected\": 3, \"passed\": false}], \"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."}}