FA-100821 / Model output stopping rules / Open access

Output stopping: A full token budget is not reported as a length finish · case 01

A full token budget is not reported as a length finish.

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

ROOT CAUSE

A strict comparison misses output that fills the configured token budget exactly.

THE FAILURE

A strict comparison misses output that fills the configured token budget exactly.

Unsuccessful approach: Testing equality alone handles an exact fill but misses an already-over-budget count.

Case contract

Return length when the emitted token count has reached or exceeded the nonnegative cap; otherwise return None. This includes cap zero with an empty output.

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, cap):
    return 'length' if len(tokens)>cap else None
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('exact cap',solve(list(range(N)),N),'length')
check('below cap',solve(list(range(max(0,N-1))),N),None)
check('over cap',solve(list(range(N+1)),N),'length')
check('zero cap empty',solve([],0),'length')
check('positive cap empty',solve([],N),None)
check('singleton cap',solve([7],1),'length')
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
exact cap	`None`	`length`	Failed
below cap	`None`	`None`	Passed
over cap	`length`	`length`	Passed
zero cap empty	`None`	`length`	Failed
positive cap empty	`None`	`None`	Passed
singleton cap	`None`	`length`	Failed

SHA-256 / d5867e34a1aef3b2d7d1d4ab2628755cc869b4f04cf778228429b50782111e3a

2 / The unsuccessful fix

Exit 1

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

N = 1
observations = []
def solve(tokens, cap):
    return 'length' if len(tokens)==cap else None
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('exact cap',solve(list(range(N)),N),'length')
check('below cap',solve(list(range(max(0,N-1))),N),None)
check('over cap',solve(list(range(N+1)),N),'length')
check('zero cap empty',solve([],0),'length')
check('positive cap empty',solve([],N),None)
check('singleton cap',solve([7],1),'length')
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
exact cap	`length`	`length`	Passed
below cap	`None`	`None`	Passed
over cap	`None`	`length`	Failed
zero cap empty	`length`	`length`	Passed
positive cap empty	`None`	`None`	Passed
singleton cap	`length`	`length`	Passed

SHA-256 / 9cafcd86df8599506836edb74624c33ca1e004c4e2244f7866325b07fffb870e

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 2026-09-30T16:48:40.054341+00:00.

Case digest / e41789123dbd93727b57a3cc82724f48df7673d043bd18a86672afab359dd4e2