FAILURE MAP
← Case archive

FA-100766 / Model output stopping rules / Open access

Output stopping: Stop matches are honored before minimum length · case 01

Stop matches are honored before minimum length.

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

ROOT CAUSE

The stop test ignores the minimum number of generated tokens.

THE FAILURE

The stop test ignores the minimum number of generated tokens.

Unsuccessful approach: A strict greater-than check still rejects a stop exactly at the allowed minimum.

Case contract

Return generated tokens before the first configured stop ID only when their zero-based position is at least min_new; earlier stop IDs are retained as ordinary output. Also return whether a stop was accepted.

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, stop_ids, min_new):
    for i,t in enumerate(tokens):
        if t in stop_ids:return tokens[:i],True
    return tokens,False
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('stop at minimum',solve([1,8,2],[8],1),([1],True))
check('too early retained',solve([8,2],[8],1),([8,2],False))
check('later stop',solve([1,2,8,3],[8],1),([1,2],True))
check('no stop',solve([1,2],[8],1),([1,2],False))
check('minimum zero',solve([8],[8],0),([],True))
check('empty output',solve([],[8],0),([],False))
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
stop at minimum[[1], True][[1], True]Passed
too early retained[[], True][[8, 2], False]Failed
later stop[[1, 2], True][[1, 2], True]Passed
no stop[[1, 2], False][[1, 2], False]Passed
minimum zero[[], True][[], True]Passed
empty output[[], False][[], False]Passed

SHA-256 / 5ef2c23cc27c50ab4a302a77df4897d665b4e6f68e2fa61b1be224eba3962a2b

2 / The unsuccessful fix

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

N = 1
observations = []
def solve(tokens, stop_ids, min_new):
    for i,t in enumerate(tokens):
        if t in stop_ids and i>min_new:return tokens[:i],True
    return tokens,False
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('stop at minimum',solve([1,8,2],[8],1),([1],True))
check('too early retained',solve([8,2],[8],1),([8,2],False))
check('later stop',solve([1,2,8,3],[8],1),([1,2],True))
check('no stop',solve([1,2],[8],1),([1,2],False))
check('minimum zero',solve([8],[8],0),([],True))
check('empty output',solve([],[8],0),([],False))
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
stop at minimum[[1, 8, 2], False][[1], True]Failed
too early retained[[8, 2], False][[8, 2], False]Passed
later stop[[1, 2], True][[1, 2], True]Passed
no stop[[1, 2], False][[1, 2], False]Passed
minimum zero[[8], False][[], True]Failed
empty output[[], False][[], False]Passed

SHA-256 / 7ac4cb111a054c359063a29312f432c559e7c9e86896b445acd5e242d59f22e8

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 2026-09-30T16:48:39.593655+00:00.

Case digest / 8ef021c378adc621bc8321cb1a743252fade4ba096da383a9f230cc643e3322c