{"abstract":"Bias parameter receives unintended weight decay.","category":"Optimizer update state","checks":6,"contract":"Records [role,p,g], lr=1, wd=.5; p_next=p-g-wd*p only for role weight.","contract_signature":"records","evaluation_group":"optimizer-toy-weight-decay-exempt-bias","failed_approach":"Removing decay from all roles repairs biases but also exempts regularized weights.","family":"w3-optimizer-update-state-weight-decay-exempt-bias","id":"FA-99031","implementations":{"attempt":{"sha256":"ee71557c8133519d5d54dacd727d40666c2c4b9a61c3fe31bf4b072c883a58eb","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(records):\n    return [[role,p-g] for role,p,g in records]\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\ncheck('weight and bias',solve([['weight',4,1],['bias',4,1]]),[['weight',1],['bias',3]])\ncheck('bias only',solve([['bias',N,2]]),[['bias',N-2]])\ncheck('weight only',solve([['weight',N,0]]),[['weight',N/2]])\ncheck('empty params',solve([]),[])\ncheck('negative weight',solve([['weight',-4,0]]),[['weight',-2]])\ncheck('negative bias',solve([['bias',-4,1]]),[['bias',-5]])\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":"382a4ba9495639e17081021b4fe224c9792d13f3e099301b45613200c7ef8652","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(records):\n    return [[role,p-g-0.5*p] for role,p,g in records]\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\ncheck('weight and bias',solve([['weight',4,1],['bias',4,1]]),[['weight',1],['bias',3]])\ncheck('bias only',solve([['bias',N,2]]),[['bias',N-2]])\ncheck('weight only',solve([['weight',N,0]]),[['weight',N/2]])\ncheck('empty params',solve([]),[])\ncheck('negative weight',solve([['weight',-4,0]]),[['weight',-2]])\ncheck('negative bias',solve([['bias',-4,1]]),[['bias',-5]])\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":"These are deterministic scalar or short-vector toy contracts, not claims about production optimizer APIs or training quality. 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-optimizer-update-state-weight-decay-exempt-bias","generated_at":"2026-09-30T16:48:19.456541+00:00","license":"CC0-1.0","python":"3.12.14","seed":1,"split":"open-access"},"relevance":"Optimizer implementations must associate each parameter with its own mathematically specified update state.","root_cause":"All parameter roles share the same decay path.","sha256":"3af8c4ed5432006f4603a29bd483f8253cf6a87a8b05703a7bbdaeb1ed275f2d","title":"Bias parameter receives unintended weight decay · 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":27.977,"exit_code":1,"observations":[{"actual":[["weight",3],["bias",3]],"check":"weight and bias","expected":[["weight",1],["bias",3]],"passed":false},{"actual":[["bias",-1]],"check":"bias only","expected":[["bias",-1]],"passed":true},{"actual":[["weight",1]],"check":"weight only","expected":[["weight",0.5]],"passed":false},{"actual":[],"check":"empty params","expected":[],"passed":true},{"actual":[["weight",-4]],"check":"negative weight","expected":[["weight",-2]],"passed":false},{"actual":[["bias",-5]],"check":"negative bias","expected":[["bias",-5]],"passed":true}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"weight and bias\", \"actual\": [[\"weight\", 3], [\"bias\", 3]], \"expected\": [[\"weight\", 1], [\"bias\", 3]], \"passed\": false}, {\"check\": \"bias only\", \"actual\": [[\"bias\", -1]], \"expected\": [[\"bias\", -1]], \"passed\": true}, {\"check\": \"weight only\", \"actual\": [[\"weight\", 1]], \"expected\": [[\"weight\", 0.5]], \"passed\": false}, {\"check\": \"empty params\", \"actual\": [], \"expected\": [], \"passed\": true}, {\"check\": \"negative weight\", \"actual\": [[\"weight\", -4]], \"expected\": [[\"weight\", -2]], \"passed\": false}, {\"check\": \"negative bias\", \"actual\": [[\"bias\", -5]], \"expected\": [[\"bias\", -5]], \"passed\": true}], \"passed\": false}\n"},"broken":{"elapsed_ms":27.307,"exit_code":1,"observations":[{"actual":[["weight",1.0],["bias",1.0]],"check":"weight and bias","expected":[["weight",1],["bias",3]],"passed":false},{"actual":[["bias",-1.5]],"check":"bias only","expected":[["bias",-1]],"passed":false},{"actual":[["weight",0.5]],"check":"weight only","expected":[["weight",0.5]],"passed":true},{"actual":[],"check":"empty params","expected":[],"passed":true},{"actual":[["weight",-2.0]],"check":"negative weight","expected":[["weight",-2]],"passed":true},{"actual":[["bias",-3.0]],"check":"negative bias","expected":[["bias",-5]],"passed":false}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"weight and bias\", \"actual\": [[\"weight\", 1.0], [\"bias\", 1.0]], \"expected\": [[\"weight\", 1], [\"bias\", 3]], \"passed\": false}, {\"check\": \"bias only\", \"actual\": [[\"bias\", -1.5]], \"expected\": [[\"bias\", -1]], \"passed\": false}, {\"check\": \"weight only\", \"actual\": [[\"weight\", 0.5]], \"expected\": [[\"weight\", 0.5]], \"passed\": true}, {\"check\": \"empty params\", \"actual\": [], \"expected\": [], \"passed\": true}, {\"check\": \"negative weight\", \"actual\": [[\"weight\", -2.0]], \"expected\": [[\"weight\", -2]], \"passed\": true}, {\"check\": \"negative bias\", \"actual\": [[\"bias\", -3.0]], \"expected\": [[\"bias\", -5]], \"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."}}