{"abstract":"Mean-reduced microbatch losses are summed.","category":"Gradient accumulation weighting","checks":7,"contract":"Each input is one sum-reduced microbatch contribution. Return the total without dividing by a count.","contract_signature":"loss_sums","evaluation_group":"gradient-accumulation-remainder_window","failed_approach":"Counting only positive terms wrongly removes valid zero or negative contributions.","family":"w3-gradient-accumulation-weighting-sum-reduction","id":"FA-98866","implementations":{"attempt":{"sha256":"b85332e19ec6533535fa383c719f2fc480c8a725d9ca0e5498173bc93e4d137c","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(loss_sums):\n    return sum(loss_sums) / max(1,sum(s > 0 for s in loss_sums))\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\ncheck('two sums',solve([2*N,4*N]),6*N)\ncheck('unequal values',solve([2*N,12*N]),14*N)\ncheck('single',solve([5*N]),5*N)\ncheck('empty',solve([]),0)\ncheck('zero included',solve([0,8*N]),8*N)\ncheck('negative contribution',solve([3*N,-N]),2*N)\ncheck('three sums',solve([N,2*N,3*N]),6*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":"4fdc05ce11e34d722e2fd833423dfb3e7ce39920068e01ac6146b14137fa0969","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(loss_sums):\n    return sum(loss_sums) / max(1,len(loss_sums))\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\ncheck('two sums',solve([2*N,4*N]),6*N)\ncheck('unequal values',solve([2*N,12*N]),14*N)\ncheck('single',solve([5*N]),5*N)\ncheck('empty',solve([]),0)\ncheck('zero included',solve([0,8*N]),8*N)\ncheck('negative contribution',solve([3*N,-N]),2*N)\ncheck('three sums',solve([N,2*N,3*N]),6*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 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-sum-reduction","generated_at":"2026-09-30T16:48:17.605383+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 broken implementation averages sum-reduced contributions by number of microbatches.","sha256":"a5a0160dce9a9681e070ba01d15c05fffe6bce7fc242c208dc90111ab57c675d","title":"Mean-reduced microbatch losses are summed · 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":27.339,"exit_code":1,"observations":[{"actual":3.0,"check":"two sums","expected":6,"passed":false},{"actual":7.0,"check":"unequal values","expected":14,"passed":false},{"actual":5.0,"check":"single","expected":5,"passed":true},{"actual":0.0,"check":"empty","expected":0,"passed":true},{"actual":8.0,"check":"zero included","expected":8,"passed":true},{"actual":2.0,"check":"negative contribution","expected":2,"passed":true},{"actual":2.0,"check":"three sums","expected":6,"passed":false}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"two sums\", \"actual\": 3.0, \"expected\": 6, \"passed\": false}, {\"check\": \"unequal values\", \"actual\": 7.0, \"expected\": 14, \"passed\": false}, {\"check\": \"single\", \"actual\": 5.0, \"expected\": 5, \"passed\": true}, {\"check\": \"empty\", \"actual\": 0.0, \"expected\": 0, \"passed\": true}, {\"check\": \"zero included\", \"actual\": 8.0, \"expected\": 8, \"passed\": true}, {\"check\": \"negative contribution\", \"actual\": 2.0, \"expected\": 2, \"passed\": true}, {\"check\": \"three sums\", \"actual\": 2.0, \"expected\": 6, \"passed\": false}], \"passed\": false}\n"},"broken":{"elapsed_ms":28.062,"exit_code":1,"observations":[{"actual":3.0,"check":"two sums","expected":6,"passed":false},{"actual":7.0,"check":"unequal values","expected":14,"passed":false},{"actual":5.0,"check":"single","expected":5,"passed":true},{"actual":0.0,"check":"empty","expected":0,"passed":true},{"actual":4.0,"check":"zero included","expected":8,"passed":false},{"actual":1.0,"check":"negative contribution","expected":2,"passed":false},{"actual":2.0,"check":"three sums","expected":6,"passed":false}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"two sums\", \"actual\": 3.0, \"expected\": 6, \"passed\": false}, {\"check\": \"unequal values\", \"actual\": 7.0, \"expected\": 14, \"passed\": false}, {\"check\": \"single\", \"actual\": 5.0, \"expected\": 5, \"passed\": true}, {\"check\": \"empty\", \"actual\": 0.0, \"expected\": 0, \"passed\": true}, {\"check\": \"zero included\", \"actual\": 4.0, \"expected\": 8, \"passed\": false}, {\"check\": \"negative contribution\", \"actual\": 1.0, \"expected\": 2, \"passed\": false}, {\"check\": \"three sums\", \"actual\": 2.0, \"expected\": 6, \"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."}}