FA-98131 / Sequence batch packing / Open access

Flattened packed tokens retain the wrong owner indices · case 01

Flattened packed tokens retain the wrong owner indices.

Verified by executionVariant 1 · 6 checks per implementationDownload source bundle ↓JSON ↗

ROOT CAUSE

Owner IDs are shifted to one-based values and can mismatch other row-local metadata.

THE FAILURE

Owner IDs are shifted to one-based values and can mismatch other row-local metadata.

Unsuccessful approach: Dropping the final owner marker per sequence only passes for empty inputs.

Case contract

Flatten sequences in order and return parallel zero-based sequence-owner IDs for every token.

Why this case matters

This deterministic toy packing rule models a common boundary in preparing bounded training sequences.

1 / The failure

Exit 1

"""Failure Map reference implementation. Python standard library only."""
import json

N = 1
observations = []
def solve(sequences):
    return [x for s in sequences for x in s], [i+1 for i,s in enumerate(sequences) for _ in s]
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
a=[N,2]; b=[3]*N
check('ownership lengths', solve([a,b]), (a+b,[0]*2+[1]*N))
check('empty', solve([]), ([],[]))
check('empty segment', solve([[],[7]]), ([7],[1]))
check('single segment', solve([[N]*N]), ([N]*N,[0]*N))
check('three owners', solve([[1],[2,3],[4]]), ([1,2,3,4],[0,1,1,2]))
check('equal values', solve([[1],[1]]), ([1,1],[0,1]))
print(json.dumps({"observations": observations, "passed": all(x["passed"] for x in observations)}, ensure_ascii=False))
raise SystemExit(0 if all(x["passed"] for x in observations) else 1)

Boundary fixture	Actual	Expected	Outcome
ownership lengths	`[[1, 2, 3], [1, 1, 2]]`	`[[1, 2, 3], [0, 0, 1]]`	Failed
empty	`[[], []]`	`[[], []]`	Passed
empty segment	`[[7], [2]]`	`[[7], [1]]`	Failed
single segment	`[[1], [1]]`	`[[1], [0]]`	Failed
three owners	`[[1, 2, 3, 4], [1, 2, 2, 3]]`	`[[1, 2, 3, 4], [0, 1, 1, 2]]`	Failed
equal values	`[[1, 1], [1, 2]]`	`[[1, 1], [0, 1]]`	Failed

SHA-256 / a13207a23677af110145e36acd7f6c704c0f3185b07702a6f24502984a808468

2 / The unsuccessful fix

Exit 1

"""Failure Map reference implementation. Python standard library only."""
import json

N = 1
observations = []
def solve(sequences):
    return [x for s in sequences for x in s], [i for i,s in enumerate(sequences) for _ in range(max(0,len(s)-1))]
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
a=[N,2]; b=[3]*N
check('ownership lengths', solve([a,b]), (a+b,[0]*2+[1]*N))
check('empty', solve([]), ([],[]))
check('empty segment', solve([[],[7]]), ([7],[1]))
check('single segment', solve([[N]*N]), ([N]*N,[0]*N))
check('three owners', solve([[1],[2,3],[4]]), ([1,2,3,4],[0,1,1,2]))
check('equal values', solve([[1],[1]]), ([1,1],[0,1]))
print(json.dumps({"observations": observations, "passed": all(x["passed"] for x in observations)}, ensure_ascii=False))
raise SystemExit(0 if all(x["passed"] for x in observations) else 1)

Boundary fixture	Actual	Expected	Outcome
ownership lengths	`[[1, 2, 3], [0]]`	`[[1, 2, 3], [0, 0, 1]]`	Failed
empty	`[[], []]`	`[[], []]`	Passed
empty segment	`[[7], []]`	`[[7], [1]]`	Failed
single segment	`[[1], []]`	`[[1], [0]]`	Failed
three owners	`[[1, 2, 3, 4], [1]]`	`[[1, 2, 3, 4], [0, 1, 1, 2]]`	Failed
equal values	`[[1, 1], []]`	`[[1, 1], [0, 1]]`	Failed

SHA-256 / 2bb49053f22b86cc16a51a274d8bd386fa481e1477e3da473120469472294e32

HELD IN THE MEMBER ARCHIVE

The verified repair and its recorded checks are member-only.

This mechanism has 6 recorded checks per implementation. The open-access tier publishes the failure and the unsuccessful fix; the repaired source that passes every check, and the observations that prove it, are available to members.

Every case sharing this mechanism uses the same contract and the same repair, so this one record is held back for all of them.

Member access is invitation-based. Sign in with your invited account to inspect the repair.

Verification & scope

Inputs are short integer-token lists under the stated toy contract; results do not claim compatibility with a specific training framework. 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.

Observations recorded using Python 3.12.14 at 2026-09-30T16:48:09.757740+00:00.

Case digest / 5a1c7a9f7c79cf7f5166b83a498e637c69326c3be3abfcff0ea7aab560d23c26