FA-98161 / Attention mask construction / Open access

Attention crosses independent block boundaries · case 01

Attention crosses independent block boundaries.

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

ROOT CAUSE

Visibility uses global causal order without checking block ownership.

THE FAILURE

Visibility uses global causal order without checking block ownership.

Unsuccessful approach: Restricting matches to neighboring positions removes valid long-range visibility inside a block.

Case contract

Square causal matrix; positions attend only within identical supplied block IDs.

Why this case matters

A small explicit visibility matrix makes sequence-attention boundary behavior reviewable.

1 / The failure

Exit 1

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

N = 1
observations = []
def solve(blocks):
    return [[j<=i for j in range(len(blocks))] for i in range(len(blocks))]
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
b=[0]*N+[1,1]
check('boundary isolation',solve(b),[[b[i]==b[j] and j<=i for j in range(len(b))] for i in range(len(b))])
check('same block',solve([4,4]),[[True,False],[True,True]])
check('alternating labels',solve([1,2,1]),[[True,False,False],[False,True,False],[True,False,True]])
check('empty',solve([]),[])
check('singleton',solve([7]),[[True]])
check('equal-sized block IDs differ',solve([2,3,2]),[[True,False,False],[False,True,False],[True,False,True]])
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
boundary isolation	`[[True, False, False], [True, True, False], [True, True, True]]`	`[[True, False, False], [False, True, False], [False, True, True]]`	Failed
same block	`[[True, False], [True, True]]`	`[[True, False], [True, True]]`	Passed
alternating labels	`[[True, False, False], [True, True, False], [True, True, True]]`	`[[True, False, False], [False, True, False], [True, False, True]]`	Failed
empty	`[]`	`[]`	Passed
singleton	`[[True]]`	`[[True]]`	Passed
equal-sized block IDs differ	`[[True, False, False], [True, True, False], [True, True, True]]`	`[[True, False, False], [False, True, False], [True, False, True]]`	Failed

SHA-256 / 04b18d72c2265c2ae62f0f40a5f6f887287f6e19448c05272499c0651621e641

2 / The unsuccessful fix

Exit 1

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

N = 1
observations = []
def solve(blocks):
    return [[blocks[i]==blocks[j] and j<=i and i-j<=1 for j in range(len(blocks))] for i in range(len(blocks))]
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
b=[0]*N+[1,1]
check('boundary isolation',solve(b),[[b[i]==b[j] and j<=i for j in range(len(b))] for i in range(len(b))])
check('same block',solve([4,4]),[[True,False],[True,True]])
check('alternating labels',solve([1,2,1]),[[True,False,False],[False,True,False],[True,False,True]])
check('empty',solve([]),[])
check('singleton',solve([7]),[[True]])
check('equal-sized block IDs differ',solve([2,3,2]),[[True,False,False],[False,True,False],[True,False,True]])
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
boundary isolation	`[[True, False, False], [False, True, False], [False, True, True]]`	`[[True, False, False], [False, True, False], [False, True, True]]`	Passed
same block	`[[True, False], [True, True]]`	`[[True, False], [True, True]]`	Passed
alternating labels	`[[True, False, False], [False, True, False], [False, False, True]]`	`[[True, False, False], [False, True, False], [True, False, True]]`	Failed
empty	`[]`	`[]`	Passed
singleton	`[[True]]`	`[[True]]`	Passed
equal-sized block IDs differ	`[[True, False, False], [False, True, False], [False, False, True]]`	`[[True, False, False], [False, True, False], [True, False, True]]`	Failed

SHA-256 / c80540fa838fb9f7fe53fbc19f0b9e92e3f720f3f6412ccdd20c0a0f6b921aed

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

These are toy Boolean matrix contracts; they do not define a production model architecture or tensor library behavior. 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 2026-09-30T16:48:10.192954+00:00.

Case digest / c83355fa874e5556ebaf9bfd75f89f51a422ebebcf375e36a465d99313e6d605