{"abstract":"A context window predicts each token from itself.","category":"Autoregressive target shifting","checks":6,"contract":"For tokens and start, return adjacent pairs whose input indexes are start through len(tokens)-2; negative start clamps to zero.","contract_signature":"tokens, start","evaluation_group":"w3-ars-context_window","failed_approach":"Adding a lookahead to the target slice but retaining the old input start skips a context token.","family":"w3-autoregressive-target-shifting-causal-offset-window","id":"FA-98281","implementations":{"attempt":{"sha256":"e5e5f1ba5a26b87d498ebfd8606a95d578f06846121421c6c5e54545fce29d9f","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(tokens, start):\n    return [[tokens[i+1],tokens[i+2]] for i in range(max(0,start),len(tokens)-2)]\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\ncheck('middle window',solve([0,1,2,3,4],2),[[2,3],[3,4]])\ncheck('start zero',solve([N,N+1,N+2],0),[[N,N+1],[N+1,N+2]])\ncheck('negative clamps',solve([1,2,3],-4),[[1,2],[2,3]])\ncheck('last pair',solve([1,2,3,4],2),[[3,4]])\ncheck('past end',solve([1,2,3],8),[])\ncheck('short',solve([1],0),[])\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":"942b99cb10e095742a18c2701d6da93c52a5e84c7d5a97b8c5d17cb425b92514","source":"\"\"\"Failure Map reference implementation. Python standard library only.\"\"\"\nimport json\n\nN = 1\nobservations = []\ndef solve(tokens, start):\n    return [[tokens[i],tokens[i]] for i in range(max(0,start),len(tokens)-1)]\ndef check(label, actual, expected):\n    observations.append({\"check\": label, \"actual\": actual, \"expected\": expected, \"passed\": actual == expected})\ncheck('middle window',solve([0,1,2,3,4],2),[[2,3],[3,4]])\ncheck('start zero',solve([N,N+1,N+2],0),[[N,N+1],[N+1,N+2]])\ncheck('negative clamps',solve([1,2,3],-4),[[1,2],[2,3]])\ncheck('last pair',solve([1,2,3,4],2),[[3,4]])\ncheck('past end',solve([1,2,3],8),[])\ncheck('short',solve([1],0),[])\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-causal-offset-window","generated_at":"2026-09-30T16:48:11.500501+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 target index is copied from the input index instead of applying the one-token causal offset.","sha256":"0998609399371d49722d6e0cc7f73861b99b54f29931722b3a0ffd85cde7db2b","title":"A context window predicts each token from itself · 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":28.685,"exit_code":1,"observations":[{"actual":[[3,4]],"check":"middle window","expected":[[2,3],[3,4]],"passed":false},{"actual":[[2,3]],"check":"start zero","expected":[[1,2],[2,3]],"passed":false},{"actual":[[2,3]],"check":"negative clamps","expected":[[1,2],[2,3]],"passed":false},{"actual":[],"check":"last pair","expected":[[3,4]],"passed":false},{"actual":[],"check":"past end","expected":[],"passed":true},{"actual":[],"check":"short","expected":[],"passed":true}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"middle window\", \"actual\": [[3, 4]], \"expected\": [[2, 3], [3, 4]], \"passed\": false}, {\"check\": \"start zero\", \"actual\": [[2, 3]], \"expected\": [[1, 2], [2, 3]], \"passed\": false}, {\"check\": \"negative clamps\", \"actual\": [[2, 3]], \"expected\": [[1, 2], [2, 3]], \"passed\": false}, {\"check\": \"last pair\", \"actual\": [], \"expected\": [[3, 4]], \"passed\": false}, {\"check\": \"past end\", \"actual\": [], \"expected\": [], \"passed\": true}, {\"check\": \"short\", \"actual\": [], \"expected\": [], \"passed\": true}], \"passed\": false}\n"},"broken":{"elapsed_ms":28.745,"exit_code":1,"observations":[{"actual":[[2,2],[3,3]],"check":"middle window","expected":[[2,3],[3,4]],"passed":false},{"actual":[[1,1],[2,2]],"check":"start zero","expected":[[1,2],[2,3]],"passed":false},{"actual":[[1,1],[2,2]],"check":"negative clamps","expected":[[1,2],[2,3]],"passed":false},{"actual":[[3,3]],"check":"last pair","expected":[[3,4]],"passed":false},{"actual":[],"check":"past end","expected":[],"passed":true},{"actual":[],"check":"short","expected":[],"passed":true}],"passed":false,"stderr":"","stdout":"{\"observations\": [{\"check\": \"middle window\", \"actual\": [[2, 2], [3, 3]], \"expected\": [[2, 3], [3, 4]], \"passed\": false}, {\"check\": \"start zero\", \"actual\": [[1, 1], [2, 2]], \"expected\": [[1, 2], [2, 3]], \"passed\": false}, {\"check\": \"negative clamps\", \"actual\": [[1, 1], [2, 2]], \"expected\": [[1, 2], [2, 3]], \"passed\": false}, {\"check\": \"last pair\", \"actual\": [[3, 3]], \"expected\": [[3, 4]], \"passed\": false}, {\"check\": \"past end\", \"actual\": [], \"expected\": [], \"passed\": true}, {\"check\": \"short\", \"actual\": [], \"expected\": [], \"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."}}