FA-99146 / Checkpoint tensor mapping / Open access

Checkpoint mapping: permuted dimensions pass exact shape matching · case 01

permuted dimensions pass exact shape matching.

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

ROOT CAUSE

Sorting dimensions or comparing dimension sums loses axis identity.

THE FAILURE

Sorting dimensions or comparing dimension sums loses axis identity.

Unsuccessful approach: Matching rank and equal dimension sums still accepts different per-axis extents.

Case contract

Return sorted source names whose declared shape exactly equals the target shape.

Why this case matters

A bounded offline model of checkpoint tensor bookkeeping in a training workflow.

1 / The failure

Exit 1

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

N = 1
observations = []
def solve(source, target):
    return sorted(k for k,v in source.items() if sorted(v)==sorted(target))
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('order', solve(*({'a': [3, 2], 'b': [2, 3]}, [3, 2])), ['a'])
check('equal', solve(*({'a': [3]}, [3])), ['a'])
check('rank', solve(*({'a': [3, 1]}, [3])), [])
check('empty', solve(*({}, [3])), [])
check('zero', solve(*({'a': [0, 3]}, [0, 3])), ['a'])
check('same product', solve(*({'a': [3, 2]}, [4, 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
order	`['a', 'b']`	`['a']`	Failed
equal	`['a']`	`['a']`	Passed
rank	`[]`	`[]`	Passed
empty	`[]`	`[]`	Passed
zero	`['a']`	`['a']`	Passed
same product	`[]`	`[]`	Passed

SHA-256 / b1ac9944d7b19365cc081c9843cdd9dc33f37ba3f72bbb8126cc9a562a5fcc00

2 / The unsuccessful fix

Exit 1

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

N = 1
observations = []
def solve(source, target):
    return sorted(k for k,v in source.items() if len(v)==len(target) and sum(v)==sum(target))
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('order', solve(*({'a': [3, 2], 'b': [2, 3]}, [3, 2])), ['a'])
check('equal', solve(*({'a': [3]}, [3])), ['a'])
check('rank', solve(*({'a': [3, 1]}, [3])), [])
check('empty', solve(*({}, [3])), [])
check('zero', solve(*({'a': [0, 3]}, [0, 3])), ['a'])
check('same product', solve(*({'a': [3, 2]}, [4, 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
order	`['a', 'b']`	`['a']`	Failed
equal	`['a']`	`['a']`	Passed
rank	`[]`	`[]`	Passed
empty	`[]`	`[]`	Passed
zero	`['a']`	`['a']`	Passed
same product	`['a']`	`[]`	Failed

SHA-256 / a7de4369f4013781b0466c8e7597e81a22513cc701c569baef1a1c8b857721b7

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

Inputs are already decoded Python values; this toy contract makes no claim about a framework, file format, or production checkpoint. 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 2026-09-30T16:48:20.799135+00:00.

Case digest / ccaa1d0140d871d3841847202148f9e359679ddddbc139705dbe7fb5caee5374