{"abstract":"Adam first moment drops its exponential history.","category":"Optimizer update state","checks":6,"contract":"m_next=b1*m+(1-b1)*g; return the updated first moment.","contract_signature":"g, m, b1","evaluation_group":"optimizer-toy-adam-first-moment","failed_approach":"Blending the absolute gradient retains history but reverses the contribution of negative gradients.","family":"w3-optimizer-update-state-adam-first-moment","id":"FA-98966","implementations":{"attempt":{"sha256":"b2bb0a1959d084c56b5681a1d7445c7e9a28a2abba4c8d2d1c44a35687abc12c","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(g, m, b1):\n    return b1*m+(1-b1)*abs(g)\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\ncheck('first observation',solve(4,0,0.5),2)\ncheck('retained history',solve(4,6,0.5),5)\ncheck('negative gradient',solve(-4,2,0.5),-1)\ncheck('coefficient zero',solve(3,9,0),3)\ncheck('coefficient one',solve(3,9,1),9)\ncheck('zero gradient decay',solve(0,8,0.5),4)\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":"1a768d929c7266524f62fd7242e24bfbde957523c260110bbd6335721a4deb25","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(g, m, b1):\n    return g\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\ncheck('first observation',solve(4,0,0.5),2)\ncheck('retained history',solve(4,6,0.5),5)\ncheck('negative gradient',solve(-4,2,0.5),-1)\ncheck('coefficient zero',solve(3,9,0),3)\ncheck('coefficient one',solve(3,9,1),9)\ncheck('zero gradient decay',solve(0,8,0.5),4)\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-adam-first-moment","generated_at":"2026-09-30T16:48:18.999693+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":"The first moment is set to the current gradient each step.","sha256":"73e7dd20068671e3276105a362043f4534d79110a0d1c74e6269f1502fd62538","title":"Adam first moment drops its exponential history · 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":30.94,"exit_code":1,"observations":[{"actual":2.0,"check":"first observation","expected":2,"passed":true},{"actual":5.0,"check":"retained history","expected":5,"passed":true},{"actual":3.0,"check":"negative gradient","expected":-1,"passed":false},{"actual":3,"check":"coefficient zero","expected":3,"passed":true},{"actual":9,"check":"coefficient one","expected":9,"passed":true},{"actual":4.0,"check":"zero gradient decay","expected":4,"passed":true}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"first observation\", \"actual\": 2.0, \"expected\": 2, \"passed\": true}, {\"check\": \"retained history\", \"actual\": 5.0, \"expected\": 5, \"passed\": true}, {\"check\": \"negative gradient\", \"actual\": 3.0, \"expected\": -1, \"passed\": false}, {\"check\": \"coefficient zero\", \"actual\": 3, \"expected\": 3, \"passed\": true}, {\"check\": \"coefficient one\", \"actual\": 9, \"expected\": 9, \"passed\": true}, {\"check\": \"zero gradient decay\", \"actual\": 4.0, \"expected\": 4, \"passed\": true}], \"passed\": false}\n"},"broken":{"elapsed_ms":33.014,"exit_code":1,"observations":[{"actual":4,"check":"first observation","expected":2,"passed":false},{"actual":4,"check":"retained history","expected":5,"passed":false},{"actual":-4,"check":"negative gradient","expected":-1,"passed":false},{"actual":3,"check":"coefficient zero","expected":3,"passed":true},{"actual":3,"check":"coefficient one","expected":9,"passed":false},{"actual":0,"check":"zero gradient decay","expected":4,"passed":false}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"first observation\", \"actual\": 4, \"expected\": 2, \"passed\": false}, {\"check\": \"retained history\", \"actual\": 4, \"expected\": 5, \"passed\": false}, {\"check\": \"negative gradient\", \"actual\": -4, \"expected\": -1, \"passed\": false}, {\"check\": \"coefficient zero\", \"actual\": 3, \"expected\": 3, \"passed\": true}, {\"check\": \"coefficient one\", \"actual\": 3, \"expected\": 9, \"passed\": false}, {\"check\": \"zero gradient decay\", \"actual\": 0, \"expected\": 4, \"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."}}