FAILURE MAP
← Case archive

FA-100831 / Model output stopping rules / Open access

Output stopping: Empty stop list incorrectly stops every output · case 01

Empty stop list incorrectly stops every output.

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

ROOT CAUSE

An empty configuration is treated as an immediate stop rather than no stopping condition.

THE FAILURE

An empty configuration is treated as an immediate stop rather than no stopping condition.

Unsuccessful approach: Checking only a sequence’s first token mistakes a partial prefix for a complete marker.

Case contract

With no configured stop sequences, return all tokens and None. Otherwise stop at the earliest sequence occurrence.

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(tokens, stops):
    return ([],0) if not stops else (tokens,None)
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('empty configuration',solve(list(range(N)),[]),(list(range(N)),None))
check('configured match',solve([1,8,2],[[8]]),([1],0))
check('configured miss',solve([8,2],[[8,9]]),([8,2],None))
check('empty output',solve([],[]),([],None))
check('empty sequence ignored',solve([1],[[]]),([1],None))
check('multi token match',solve([1,8,9],[[8,9]]),([1],0))
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 fixtureActualExpectedOutcome
empty configuration[[], 0][[0], None]Failed
configured match[[1, 8, 2], None][[1], 0]Failed
configured miss[[8, 2], None][[8, 2], None]Passed
empty output[[], 0][[], None]Failed
empty sequence ignored[[1], None][[1], None]Passed
multi token match[[1, 8, 9], None][[1], 0]Failed

SHA-256 / 2da789f4a9ba6eb8491b73c806fa9b380a3755ab0924fdbb6a30133a670b1049

2 / The unsuccessful fix

Exit 1
"""Failure Map reference implementation. Python standard library only."""
import json

N = 1
observations = []
def solve(tokens, stops):
    for s in stops:
        if s and s[0] in tokens:return tokens[:tokens.index(s[0])],0
    return tokens,None
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('empty configuration',solve(list(range(N)),[]),(list(range(N)),None))
check('configured match',solve([1,8,2],[[8]]),([1],0))
check('configured miss',solve([8,2],[[8,9]]),([8,2],None))
check('empty output',solve([],[]),([],None))
check('empty sequence ignored',solve([1],[[]]),([1],None))
check('multi token match',solve([1,8,9],[[8,9]]),([1],0))
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 fixtureActualExpectedOutcome
empty configuration[[0], None][[0], None]Passed
configured match[[1], 0][[1], 0]Passed
configured miss[[], 0][[8, 2], None]Failed
empty output[[], None][[], None]Passed
empty sequence ignored[[1], None][[1], None]Passed
multi token match[[1], 0][[1], 0]Passed

SHA-256 / 51b59bde34adf5a732ad88c63dcd6b3bc2d1907738770f2d5053a036ff1c92ac

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.

Sign in to the archive ↗

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 2026-09-30T16:48:40.055385+00:00.

Case digest / 638d4ffbfbf1f53b158a7d3e06b8dc7842a1171d596c8e400ce7927fc07e71a7