FA-99236 / Checkpoint tensor mapping / Open access

Checkpoint mapping: restoration plan ignores dependencies between aliases · case 01

restoration plan ignores dependencies between aliases.

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

ROOT CAUSE

Names are listed without respecting restoration dependencies.

THE FAILURE

Names are listed without respecting restoration dependencies.

Unsuccessful approach: Filtering to initially-ready names omits valid downstream restoration steps.

Case contract

Steps are [name,depends_on]; return names in stable topological order, choosing lexical available names; cycle returns [].

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(steps):
    return sorted(name for name,dep in steps)
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('chain', solve(*([['b', ['a']], ['a', []]],)), ['a', 'b'])
check('independent', solve(*([['z', []], ['a', []]],)), ['a', 'z'])
check('cycle', solve(*([['a', ['b']], ['b', ['a']]],)), [])
check('empty', solve(*([],)), [])
check('missing dependency', solve(*([['a', ['x']]],)), [])
check('fanout', solve(*([['c', ['a', 'b']], ['b', []], ['a', []]],)), ['a', 'b', 'c'])
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
chain	`['a', 'b']`	`['a', 'b']`	Passed
independent	`['a', 'z']`	`['a', 'z']`	Passed
cycle	`['a', 'b']`	`[]`	Failed
empty	`[]`	`[]`	Passed
missing dependency	`['a']`	`[]`	Failed
fanout	`['a', 'b', 'c']`	`['a', 'b', 'c']`	Passed

SHA-256 / 51e72b7ae8d8f04940f9e8a0d9c48e240c264510bc0dc1298cf29f23e2b13e24

2 / The unsuccessful fix

Exit 1

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

N = 1
observations = []
def solve(steps):
    return [name for name,dep in sorted(steps) if not dep]
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('chain', solve(*([['b', ['a']], ['a', []]],)), ['a', 'b'])
check('independent', solve(*([['z', []], ['a', []]],)), ['a', 'z'])
check('cycle', solve(*([['a', ['b']], ['b', ['a']]],)), [])
check('empty', solve(*([],)), [])
check('missing dependency', solve(*([['a', ['x']]],)), [])
check('fanout', solve(*([['c', ['a', 'b']], ['b', []], ['a', []]],)), ['a', 'b', 'c'])
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
chain	`['a']`	`['a', 'b']`	Failed
independent	`['a', 'z']`	`['a', 'z']`	Passed
cycle	`[]`	`[]`	Passed
empty	`[]`	`[]`	Passed
missing dependency	`[]`	`[]`	Passed
fanout	`['a', 'b']`	`['a', 'b', 'c']`	Failed

SHA-256 / 53f4d86afbeec6609ac09a0a61a31de3af3ddd7e86475c190e656e37c0b3caa3

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 2026-09-30T16:48:21.700747+00:00.

Case digest / 5dc4f32fcbd592e06044c750834fc79875f98061b879497f9ebc25d08176cda9