FA-100806 / Model output stopping rules / Open access

Output stopping: Any satisfied stopping criterion ends output · case 01

Any satisfied stopping criterion ends output.

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

ROOT CAUSE

All criterion flags must be true on a step, though the contract stops when any criterion is true.

THE FAILURE

All criterion flags must be true on a step, though the contract stops when any criterion is true.

Unsuccessful approach: Requiring multiple criteria still misses a true single criterion.

Case contract

Given same-length boolean criterion results per step, return first index where any criterion is true, else None.

Why this case matters

A deterministic output-control model isolates one finish or streaming boundary from any particular runtime.

1 / The failure

Exit 1

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

N = 1
observations = []
def solve(traces):
    for i,row in enumerate(traces):
        if all(row):return i
    return None
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('one criterion',solve([[False,False],[False,True]]),1)
check('first step',solve([[True,False]]),0)
check('all false',solve([[False,False]]*N),None)
check('empty trace',solve([]),None)
check('single criterion',solve([[False],[True]]),1)
check('later signal',solve([[False,False],[False,False],[True,False]]),2)
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
one criterion	`None`	`1`	Failed
first step	`None`	`0`	Failed
all false	`None`	`None`	Passed
empty trace	`None`	`None`	Passed
single criterion	`1`	`1`	Passed
later signal	`None`	`2`	Failed

SHA-256 / dc3f23df520e5b43b9b5d91c30466a338f67c4a8c1475ae63628eaac5e79a42c

2 / The unsuccessful fix

Exit 1

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

N = 1
observations = []
def solve(traces):
    for i,row in enumerate(traces):
        if any(row) and len(row)>1:return i
    return None
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('one criterion',solve([[False,False],[False,True]]),1)
check('first step',solve([[True,False]]),0)
check('all false',solve([[False,False]]*N),None)
check('empty trace',solve([]),None)
check('single criterion',solve([[False],[True]]),1)
check('later signal',solve([[False,False],[False,False],[True,False]]),2)
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
one criterion	`1`	`1`	Passed
first step	`0`	`0`	Passed
all false	`None`	`None`	Passed
empty trace	`None`	`None`	Passed
single criterion	`None`	`1`	Failed
later signal	`2`	`2`	Passed

SHA-256 / 27385bcaf41e945b88047d665d6444d878b2e0760b560378e67756cf90f028c9

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

This toy contract covers the supplied token records only; it does not claim compatibility with a particular model runtime or tokenizer. 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 2026-09-30T16:48:40.055063+00:00.

Case digest / 66ec547843f9d3e2043dfd58db72ac9563b52b341c23b76a45d44a29a2b5a79c