{"abstract":"Smoothed targets depend on classes observed in the batch.","category":"Training label alignment","checks":7,"contract":"For each valid class index y, emit length-C target with confidence at y and equal remaining mass across other C-1 classes. For C=1 emit [1]; invalid y emits zeros.","contract_signature":"labels, class_count, confidence","evaluation_group":"training-label-alignment-declared-vocabulary-smoothing","failed_approach":"Keeping the declared width but leaving every off-class entry at zero removes the smoothing mass.","family":"w3-training-label-alignment-label-smoothing-vocabulary-width","id":"FA-98031","implementations":{"attempt":{"sha256":"e918f384d01ef92b54a42de1a4fa959014d0e9961c3acb8c62c2be075f7837dd","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(labels, class_count, confidence):\n    return [[confidence if j==y else 0 for j in range(class_count)] for y in labels]\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\ncheck('declared classes beyond batch',solve([0],N+2,0.8),[[0.8]+[round(0.2/(N+1),12)]*(N+1)])\ncheck('two classes',solve([1],2,0.6),[ [0.4,0.6] ])\ncheck('single class',solve([0],1,0.2),[[1.0]])\ncheck('empty batch',solve([],N,0.8),[])\ncheck('invalid index',solve([N],N,0.7),[[0]*N])\ncheck('second row',solve([0,2],3,0.5),[[0.5,0.25,0.25],[0.25,0.25,0.5]])\ncheck('three class mass',solve([0],3,0.9),[[0.9,0.05,0.05]])\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":"2f887358fe60030b9bd7ca9aa2c09d21d9649a33258a31aba11a35b136eea4df","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(labels, class_count, confidence):\n    return [[confidence if j==y else (1-confidence)/(max(1,len(set(labels))-1)) for j in range(max(1,len(set(labels))))] for y in labels]\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\ncheck('declared classes beyond batch',solve([0],N+2,0.8),[[0.8]+[round(0.2/(N+1),12)]*(N+1)])\ncheck('two classes',solve([1],2,0.6),[ [0.4,0.6] ])\ncheck('single class',solve([0],1,0.2),[[1.0]])\ncheck('empty batch',solve([],N,0.8),[])\ncheck('invalid index',solve([N],N,0.7),[[0]*N])\ncheck('second row',solve([0,2],3,0.5),[[0.5,0.25,0.25],[0.25,0.25,0.5]])\ncheck('three class mass',solve([0],3,0.9),[[0.9,0.05,0.05]])\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":"Confidence is between zero and one; class_count is positive. 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-training-label-alignment-label-smoothing-vocabulary-width","generated_at":"2026-09-30T16:48:08.407445+00:00","license":"CC0-1.0","python":"3.12.14","seed":1,"split":"open-access"},"relevance":"A small deterministic training-data operation makes label ownership and ordering explicit.","root_cause":"Class width and smoothing denominator are inferred from distinct minibatch labels instead of the declared class count.","sha256":"3231575205b0cc25620e62a3ac2483ffbaec092ec799dea09d15d0cc9b286883","title":"Smoothed targets depend on classes observed in the batch · 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":27.21,"exit_code":1,"observations":[{"actual":[[0.8,0,0]],"check":"declared classes beyond batch","expected":[[0.8,0.1,0.1]],"passed":false},{"actual":[[0,0.6]],"check":"two classes","expected":[[0.4,0.6]],"passed":false},{"actual":[[0.2]],"check":"single class","expected":[[1.0]],"passed":false},{"actual":[],"check":"empty batch","expected":[],"passed":true},{"actual":[[0]],"check":"invalid index","expected":[[0]],"passed":true},{"actual":[[0.5,0,0],[0,0,0.5]],"check":"second row","expected":[[0.5,0.25,0.25],[0.25,0.25,0.5]],"passed":false},{"actual":[[0.9,0,0]],"check":"three class mass","expected":[[0.9,0.05,0.05]],"passed":false}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"declared classes beyond batch\", \"actual\": [[0.8, 0, 0]], \"expected\": [[0.8, 0.1, 0.1]], \"passed\": false}, {\"check\": \"two classes\", \"actual\": [[0, 0.6]], \"expected\": [[0.4, 0.6]], \"passed\": false}, {\"check\": \"single class\", \"actual\": [[0.2]], \"expected\": [[1.0]], \"passed\": false}, {\"check\": \"empty batch\", \"actual\": [], \"expected\": [], \"passed\": true}, {\"check\": \"invalid index\", \"actual\": [[0]], \"expected\": [[0]], \"passed\": true}, {\"check\": \"second row\", \"actual\": [[0.5, 0, 0], [0, 0, 0.5]], \"expected\": [[0.5, 0.25, 0.25], [0.25, 0.25, 0.5]], \"passed\": false}, {\"check\": \"three class mass\", \"actual\": [[0.9, 0, 0]], \"expected\": [[0.9, 0.05, 0.05]], \"passed\": false}], \"passed\": false}\n"},"broken":{"elapsed_ms":24.888,"exit_code":1,"observations":[{"actual":[[0.8]],"check":"declared classes beyond batch","expected":[[0.8,0.1,0.1]],"passed":false},{"actual":[[0.4]],"check":"two classes","expected":[[0.4,0.6]],"passed":false},{"actual":[[0.2]],"check":"single class","expected":[[1.0]],"passed":false},{"actual":[],"check":"empty batch","expected":[],"passed":true},{"actual":[[0.30000000000000004]],"check":"invalid index","expected":[[0]],"passed":false},{"actual":[[0.5,0.5],[0.5,0.5]],"check":"second row","expected":[[0.5,0.25,0.25],[0.25,0.25,0.5]],"passed":false},{"actual":[[0.9]],"check":"three class mass","expected":[[0.9,0.05,0.05]],"passed":false}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"declared classes beyond batch\", \"actual\": [[0.8]], \"expected\": [[0.8, 0.1, 0.1]], \"passed\": false}, {\"check\": \"two classes\", \"actual\": [[0.4]], \"expected\": [[0.4, 0.6]], \"passed\": false}, {\"check\": \"single class\", \"actual\": [[0.2]], \"expected\": [[1.0]], \"passed\": false}, {\"check\": \"empty batch\", \"actual\": [], \"expected\": [], \"passed\": true}, {\"check\": \"invalid index\", \"actual\": [[0.30000000000000004]], \"expected\": [[0]], \"passed\": false}, {\"check\": \"second row\", \"actual\": [[0.5, 0.5], [0.5, 0.5]], \"expected\": [[0.5, 0.25, 0.25], [0.25, 0.25, 0.5]], \"passed\": false}, {\"check\": \"three class mass\", \"actual\": [[0.9]], \"expected\": [[0.9, 0.05, 0.05]], \"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."}}