{"abstract":"Expert load counts routes rather than distinct tokens.","category":"Mixture of experts dispatch","checks":6,"contract":"Rows [token,expert] may repeat; return sorted [expert,distinct-token-count] pairs.","contract_signature":"routes","evaluation_group":"moe-expert-load-count","failed_approach":"Deduplicating expert IDs counts only one token for the whole expert.","family":"w3-mixture-of-experts-dispatch-load-count","id":"FA-100591","implementations":{"attempt":{"sha256":"64d8b27d43b5ab4358e8552eb71eab5ff14bd36693b4222254ded199e85e55aa","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(routes):\n    return [[e,1] for e in sorted({e for t,e in routes})]\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\na='a'+str(N); b='b'+str(N)\nrows=[[1,a],[1,a],[2,a],[3,b],[3,b]]\ncheck('duplicate ignored',solve(rows),[[a,2],[b,1]])\ncheck('one token',solve([[N,a],[N,a]]),[[a,1]])\ncheck('empty',solve([]),[])\ncheck('distinct tokens',solve([[1,a],[2,a],[3,a]]),[[a,3]])\ncheck('multiple experts',solve([[1,a],[1,b]]),[[a,1],[b,1]])\ncheck('N token',solve([[i,a] for i in range(N)]),[[a,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":"468f0eceb48e412071ca051618a01603505bb39764a490972135dd9f6aaed002","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(routes):\n    d={}\n    for t,e in routes: d[e]=d.get(e,0)+1\n    return sorted([e,c] for e,c in d.items())\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\na='a'+str(N); b='b'+str(N)\nrows=[[1,a],[1,a],[2,a],[3,b],[3,b]]\ncheck('duplicate ignored',solve(rows),[[a,2],[b,1]])\ncheck('one token',solve([[N,a],[N,a]]),[[a,1]])\ncheck('empty',solve([]),[])\ncheck('distinct tokens',solve([[1,a],[2,a],[3,a]]),[[a,3]])\ncheck('multiple experts',solve([[1,a],[1,b]]),[[a,1],[b,1]])\ncheck('N token',solve([[i,a] for i in range(N)]),[[a,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 finite toy scores and routes; the contract does not model a production model, accelerator, or learned routing 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-mixture-of-experts-dispatch-load-count","generated_at":"2026-09-30T16:48:37.194350+00:00","license":"CC0-1.0","python":"3.12.14","seed":1,"split":"open-access"},"relevance":"A deterministic toy router makes this dispatch decision independently inspectable.","root_cause":"Duplicate route records inflate per-expert load.","sha256":"33518889bc0530f88917ef99056dd108e571674cc4b790861fc100676466b16d","title":"Expert load counts routes rather than distinct tokens · 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":31.594,"exit_code":1,"observations":[{"actual":[["a1",1],["b1",1]],"check":"duplicate ignored","expected":[["a1",2],["b1",1]],"passed":false},{"actual":[["a1",1]],"check":"one token","expected":[["a1",1]],"passed":true},{"actual":[],"check":"empty","expected":[],"passed":true},{"actual":[["a1",1]],"check":"distinct tokens","expected":[["a1",3]],"passed":false},{"actual":[["a1",1],["b1",1]],"check":"multiple experts","expected":[["a1",1],["b1",1]],"passed":true},{"actual":[["a1",1]],"check":"N token","expected":[["a1",1]],"passed":true}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"duplicate ignored\", \"actual\": [[\"a1\", 1], [\"b1\", 1]], \"expected\": [[\"a1\", 2], [\"b1\", 1]], \"passed\": false}, {\"check\": \"one token\", \"actual\": [[\"a1\", 1]], \"expected\": [[\"a1\", 1]], \"passed\": true}, {\"check\": \"empty\", \"actual\": [], \"expected\": [], \"passed\": true}, {\"check\": \"distinct tokens\", \"actual\": [[\"a1\", 1]], \"expected\": [[\"a1\", 3]], \"passed\": false}, {\"check\": \"multiple experts\", \"actual\": [[\"a1\", 1], [\"b1\", 1]], \"expected\": [[\"a1\", 1], [\"b1\", 1]], \"passed\": true}, {\"check\": \"N token\", \"actual\": [[\"a1\", 1]], \"expected\": [[\"a1\", 1]], \"passed\": true}], \"passed\": false}\n"},"broken":{"elapsed_ms":29.925,"exit_code":1,"observations":[{"actual":[["a1",3],["b1",2]],"check":"duplicate ignored","expected":[["a1",2],["b1",1]],"passed":false},{"actual":[["a1",2]],"check":"one token","expected":[["a1",1]],"passed":false},{"actual":[],"check":"empty","expected":[],"passed":true},{"actual":[["a1",3]],"check":"distinct tokens","expected":[["a1",3]],"passed":true},{"actual":[["a1",1],["b1",1]],"check":"multiple experts","expected":[["a1",1],["b1",1]],"passed":true},{"actual":[["a1",1]],"check":"N token","expected":[["a1",1]],"passed":true}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"duplicate ignored\", \"actual\": [[\"a1\", 3], [\"b1\", 2]], \"expected\": [[\"a1\", 2], [\"b1\", 1]], \"passed\": false}, {\"check\": \"one token\", \"actual\": [[\"a1\", 2]], \"expected\": [[\"a1\", 1]], \"passed\": false}, {\"check\": \"empty\", \"actual\": [], \"expected\": [], \"passed\": true}, {\"check\": \"distinct tokens\", \"actual\": [[\"a1\", 3]], \"expected\": [[\"a1\", 3]], \"passed\": true}, {\"check\": \"multiple experts\", \"actual\": [[\"a1\", 1], [\"b1\", 1]], \"expected\": [[\"a1\", 1], [\"b1\", 1]], \"passed\": true}, {\"check\": \"N token\", \"actual\": [[\"a1\", 1]], \"expected\": [[\"a1\", 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."}}