{"abstract":"Sequence means are averaged for a token-mean objective.","category":"Gradient accumulation weighting","checks":7,"contract":"Aggregate sequence loss sums and token counts, then divide once by total tokens. No tokens yields zero.","contract_signature":"sequence_sums, token_counts","evaluation_group":"gradient-accumulation-token_objective","failed_approach":"Averaging means still overweights short sequences.","family":"w3-gradient-accumulation-weighting-token-objective","id":"FA-98876","implementations":{"attempt":{"sha256":"4f4f75b8f347e62aacc3565e92d9712082237e7730baba07e3cae8f616a2c3c4","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(sequence_sums, token_counts):\n    return sum(s / c for s,c in zip(sequence_sums,token_counts) if c) / max(1,len(sequence_sums))\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\ncheck('different lengths',solve([2*N,12*N],[N,3*N]),3.5)\ncheck('equal lengths',solve([2*N,4*N],[N,N]),3)\ncheck('one sequence',solve([6*N],[2*N]),3)\ncheck('empty',solve([],[]),0)\ncheck('empty sequence ignored',solve([0,8*N],[0,2*N]),4)\ncheck('zero objective',solve([0,9*N],[N,3*N]),2.25)\ncheck('three sequences',solve([N,6*N,8*N],[1,2,4]),15*N/7)\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":"bfb702839dd418b91654e97209ba3741da5369c5d84a276d70a2865e0182d45e","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(sequence_sums, token_counts):\n    return sum(s / c for s,c in zip(sequence_sums,token_counts) if c) / max(1,sum(c > 0 for c in token_counts))\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\ncheck('different lengths',solve([2*N,12*N],[N,3*N]),3.5)\ncheck('equal lengths',solve([2*N,4*N],[N,N]),3)\ncheck('one sequence',solve([6*N],[2*N]),3)\ncheck('empty',solve([],[]),0)\ncheck('empty sequence ignored',solve([0,8*N],[0,2*N]),4)\ncheck('zero objective',solve([0,9*N],[N,3*N]),2.25)\ncheck('three sequences',solve([N,6*N,8*N],[1,2,4]),15*N/7)\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-token-objective","generated_at":"2026-09-30T16:48:17.605611+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 path averages per-sequence means, giving short and long sequences equal weight.","sha256":"a82b566e0a9de5883bc2eb474446db27257e9f97d27d5323fa4ce5d5abff039e","title":"Sequence means are averaged for a token-mean objective · 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":26.157,"exit_code":1,"observations":[{"actual":3.0,"check":"different lengths","expected":3.5,"passed":false},{"actual":3.0,"check":"equal lengths","expected":3,"passed":true},{"actual":3.0,"check":"one sequence","expected":3,"passed":true},{"actual":0.0,"check":"empty","expected":0,"passed":true},{"actual":2.0,"check":"empty sequence ignored","expected":4,"passed":false},{"actual":1.5,"check":"zero objective","expected":2.25,"passed":false},{"actual":2.0,"check":"three sequences","expected":2.142857142857143,"passed":false}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"different lengths\", \"actual\": 3.0, \"expected\": 3.5, \"passed\": false}, {\"check\": \"equal lengths\", \"actual\": 3.0, \"expected\": 3, \"passed\": true}, {\"check\": \"one sequence\", \"actual\": 3.0, \"expected\": 3, \"passed\": true}, {\"check\": \"empty\", \"actual\": 0.0, \"expected\": 0, \"passed\": true}, {\"check\": \"empty sequence ignored\", \"actual\": 2.0, \"expected\": 4, \"passed\": false}, {\"check\": \"zero objective\", \"actual\": 1.5, \"expected\": 2.25, \"passed\": false}, {\"check\": \"three sequences\", \"actual\": 2.0, \"expected\": 2.142857142857143, \"passed\": false}], \"passed\": false}\n"},"broken":{"elapsed_ms":26.887,"exit_code":1,"observations":[{"actual":3.0,"check":"different lengths","expected":3.5,"passed":false},{"actual":3.0,"check":"equal lengths","expected":3,"passed":true},{"actual":3.0,"check":"one sequence","expected":3,"passed":true},{"actual":0.0,"check":"empty","expected":0,"passed":true},{"actual":4.0,"check":"empty sequence ignored","expected":4,"passed":true},{"actual":1.5,"check":"zero objective","expected":2.25,"passed":false},{"actual":2.0,"check":"three sequences","expected":2.142857142857143,"passed":false}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"different lengths\", \"actual\": 3.0, \"expected\": 3.5, \"passed\": false}, {\"check\": \"equal lengths\", \"actual\": 3.0, \"expected\": 3, \"passed\": true}, {\"check\": \"one sequence\", \"actual\": 3.0, \"expected\": 3, \"passed\": true}, {\"check\": \"empty\", \"actual\": 0.0, \"expected\": 0, \"passed\": true}, {\"check\": \"empty sequence ignored\", \"actual\": 4.0, \"expected\": 4, \"passed\": true}, {\"check\": \"zero objective\", \"actual\": 1.5, \"expected\": 2.25, \"passed\": false}, {\"check\": \"three sequences\", \"actual\": 2.0, \"expected\": 2.142857142857143, \"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."}}