{"abstract":"Ignored targets remain in the loss denominator.","category":"Autoregressive target shifting","checks":6,"contract":"Count true mask entries whose indexes correspond to labels[1:]; absent mask positions count as false.","contract_signature":"labels, mask","evaluation_group":"w3-ars-loss_count","failed_approach":"Subtracting one from the number of valid labels fails when ignore markers occur in the context.","family":"w3-autoregressive-target-shifting-loss-count-empty-target","id":"FA-98316","implementations":{"attempt":{"sha256":"c015f2f63a84f552ff40e32e833761d66f64ead16d2c196b6d1fb69112d378b3","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(labels, mask):\n    return max(0,sum(x is not None for x in labels)-1)\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\ncheck('context only',solve([9],[]),0)\ncheck('one target',solve([9,4],[1]),1)\ncheck('ignored context',solve([None,4,5],[1,1]),2)\ncheck('ignored target mask',solve([9,4,5],[0,1]),1)\ncheck('empty',solve([],[]),0)\ncheck('mask excess',solve([1,2],[1,1,1]),1)\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":"0180e61b18b7aff3b89f74e6862faf275d6a6c354a98bba901ae76bdc61619e7","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(labels, mask):\n    return max(0,len(labels)-1)\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\ncheck('context only',solve([9],[]),0)\ncheck('one target',solve([9,4],[1]),1)\ncheck('ignored context',solve([None,4,5],[1,1]),2)\ncheck('ignored target mask',solve([9,4,5],[0,1]),1)\ncheck('empty',solve([],[]),0)\ncheck('mask excess',solve([1,2],[1,1,1]),1)\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 toy integer token sequences; this contract says nothing about model quality, batching, sampling, or production training frameworks. 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-autoregressive-target-shifting-loss-count-empty-target","generated_at":"2026-09-30T16:48:11.499935+00:00","license":"CC0-1.0","python":"3.12.14","seed":1,"split":"open-access"},"relevance":"A small deterministic token-stream contract makes next-token alignment errors inspectable.","root_cause":"The denominator is inferred from sequence length without applying the target supervision mask.","sha256":"54cb9323cb72000ea592dde40e0b2a0f4ddd004e91be4455bac90b871b575d7b","title":"Ignored targets remain in the loss denominator · case 01","variant":1,"variant_policy":"The same explicitly authored fixtures repeat across all five numbered records; runtime N is unused. These records share one mechanism and must remain together in evaluation splits.","verified":true,"visibility":"public","verification":{"attempt":{"elapsed_ms":28.384,"exit_code":1,"observations":[{"actual":0,"check":"context only","expected":0,"passed":true},{"actual":1,"check":"one target","expected":1,"passed":true},{"actual":1,"check":"ignored context","expected":2,"passed":false},{"actual":2,"check":"ignored target mask","expected":1,"passed":false},{"actual":0,"check":"empty","expected":0,"passed":true},{"actual":1,"check":"mask excess","expected":1,"passed":true}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"context only\", \"actual\": 0, \"expected\": 0, \"passed\": true}, {\"check\": \"one target\", \"actual\": 1, \"expected\": 1, \"passed\": true}, {\"check\": \"ignored context\", \"actual\": 1, \"expected\": 2, \"passed\": false}, {\"check\": \"ignored target mask\", \"actual\": 2, \"expected\": 1, \"passed\": false}, {\"check\": \"empty\", \"actual\": 0, \"expected\": 0, \"passed\": true}, {\"check\": \"mask excess\", \"actual\": 1, \"expected\": 1, \"passed\": true}], \"passed\": false}\n"},"broken":{"elapsed_ms":27.772,"exit_code":1,"observations":[{"actual":0,"check":"context only","expected":0,"passed":true},{"actual":1,"check":"one target","expected":1,"passed":true},{"actual":2,"check":"ignored context","expected":2,"passed":true},{"actual":2,"check":"ignored target mask","expected":1,"passed":false},{"actual":0,"check":"empty","expected":0,"passed":true},{"actual":1,"check":"mask excess","expected":1,"passed":true}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"context only\", \"actual\": 0, \"expected\": 0, \"passed\": true}, {\"check\": \"one target\", \"actual\": 1, \"expected\": 1, \"passed\": true}, {\"check\": \"ignored context\", \"actual\": 2, \"expected\": 2, \"passed\": true}, {\"check\": \"ignored target mask\", \"actual\": 2, \"expected\": 1, \"passed\": false}, {\"check\": \"empty\", \"actual\": 0, \"expected\": 0, \"passed\": true}, {\"check\": \"mask excess\", \"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."}}