FA-100641 / Speculative decoding bookkeeping / Open access
Decoding: Acceptance count includes tokens after the first mismatch · case 01
Acceptance count includes tokens after the first mismatch.
ROOT CAUSE
The count uses total equal positions rather than stopping at the first rejection.
THE FAILURE
The count uses total equal positions rather than stopping at the first rejection.
Unsuccessful approach: Comparing only the first token fixes immediate rejection but misses later mismatch positions.
Case contract
Given equal-length draft and verified token lists, return the length of their common prefix.
Why this case matters
A self-contained bookkeeping model makes prefix and token-count semantics observable without a model runtime.
1 / The failure
Exit 1"""Failure Map reference implementation. Python standard library only."""
import json
N = 1
observations = []
def solve(draft, verified):
return sum(a == b for a,b in zip(draft,verified))
def check(label, actual, expected):
observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('full match', solve([1,2,N],[1,2,N]),3)
check('middle rejection', solve([1,2,3],[1,9,3]),1)
check('first rejection', solve([1],[2]),0)
check('empty draft', solve([],[]),0)
check('short verified', solve([1,2],[1]),1)
check('empty verified', solve([1],[]),0)
check('single match', solve([N],[N]),1)
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 |
|---|---|---|---|
| full match | 3 | 3 | Passed |
| middle rejection | 2 | 1 | Failed |
| first rejection | 0 | 0 | Passed |
| empty draft | 0 | 0 | Passed |
| short verified | 1 | 1 | Passed |
| empty verified | 0 | 0 | Passed |
| single match | 1 | 1 | Passed |
SHA-256 / bc99fb29ea61f6b43fa88f93e5fc932ee8e87b3da3d2aa7b7c07a20a27e22acd
2 / The unsuccessful fix
Exit 1"""Failure Map reference implementation. Python standard library only."""
import json
N = 1
observations = []
def solve(draft, verified):
return int(bool(draft) and bool(verified) and draft[0] == verified[0])
def check(label, actual, expected):
observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('full match', solve([1,2,N],[1,2,N]),3)
check('middle rejection', solve([1,2,3],[1,9,3]),1)
check('first rejection', solve([1],[2]),0)
check('empty draft', solve([],[]),0)
check('short verified', solve([1,2],[1]),1)
check('empty verified', solve([1],[]),0)
check('single match', solve([N],[N]),1)
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 |
|---|---|---|---|
| full match | 1 | 3 | Failed |
| middle rejection | 1 | 1 | Passed |
| first rejection | 0 | 0 | Passed |
| empty draft | 0 | 0 | Passed |
| short verified | 1 | 1 | Passed |
| empty verified | 0 | 0 | Passed |
| single match | 1 | 1 | Passed |
SHA-256 / a58d771fc1acfe6ab6ce0acc4048b0cba0f6b775fefca3018b43d2e3cfd0e456
HELD IN THE MEMBER ARCHIVE
The verified repair and its recorded checks are member-only.
This mechanism has 7 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
These are finite toy sequences with explicit contracts; they make no claim about production decoder behavior or model quality. 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 2026-09-30T16:48:38.067804+00:00.
Case digest / 1e78447a6808f73bc14b339a481a573cda8505b390b3ad4ff62d43866396b713