FA-99221 / Checkpoint tensor mapping / Open access

Checkpoint mapping: tied parameter aliases load conflicting values · case 01

tied parameter aliases load conflicting values.

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

ROOT CAUSE

The first present alias wins even when tied tensors disagree.

THE FAILURE

The first present alias wins even when tied tensors disagree.

Unsuccessful approach: Always reading the canonical first alias fails when only a secondary name is present.

Case contract

For alias groups, return one value per group only if all supplied members agree; absent groups are omitted.

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(aliases, incoming):
    return [next((incoming[k] for k in group if k in incoming),None) for group in aliases if any(k in incoming for k in group)]
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('conflict', solve(*([['a', 'b']], {'a': 3, 'b': 4})), [])
check('agree', solve(*([['a', 'b']], {'a': 3, 'b': 3})), [3])
check('one alias', solve(*([['a', 'b']], {'b': 3})), [3])
check('absent', solve(*([['a', 'b']], {})), [])
check('singleton', solve(*([['x']], {'x': 2})), [2])
check('two groups', solve(*([['a', 'b'], ['c', 'd']], {'a': 1, 'b': 1, 'd': 3})), [1, 3])
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
conflict	`[3]`	`[]`	Failed
agree	`[3]`	`[3]`	Passed
one alias	`[3]`	`[3]`	Passed
absent	`[]`	`[]`	Passed
singleton	`[2]`	`[2]`	Passed
two groups	`[1, 3]`	`[1, 3]`	Passed

SHA-256 / e9f2771ba814b0abc296789c9a6f10940fe5f731aade25d61c57bfc899125281

2 / The unsuccessful fix

Exit 1

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

N = 1
observations = []
def solve(aliases, incoming):
    return [incoming.get(group[0]) for group in aliases if any(k in incoming for k in group)]
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('conflict', solve(*([['a', 'b']], {'a': 3, 'b': 4})), [])
check('agree', solve(*([['a', 'b']], {'a': 3, 'b': 3})), [3])
check('one alias', solve(*([['a', 'b']], {'b': 3})), [3])
check('absent', solve(*([['a', 'b']], {})), [])
check('singleton', solve(*([['x']], {'x': 2})), [2])
check('two groups', solve(*([['a', 'b'], ['c', 'd']], {'a': 1, 'b': 1, 'd': 3})), [1, 3])
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
conflict	`[3]`	`[]`	Failed
agree	`[3]`	`[3]`	Passed
one alias	`[None]`	`[3]`	Failed
absent	`[]`	`[]`	Passed
singleton	`[2]`	`[2]`	Passed
two groups	`[1, None]`	`[1, 3]`	Failed

SHA-256 / f27d9fe67d9cfbf9fe0eb1820d21eb37ea5e2cebd8df1e4c048184c442e07682

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 2026-09-30T16:48:21.700670+00:00.

Case digest / ac9a81023dba55d2a65beaad3c74eaa5cf2d70e715d4e7078a1c21acbdbe4d6c