FA-100626 / Mixture of experts dispatch / Open access

Repeated expert contributions are not merged per token · case 01

Repeated expert contributions are not merged per token.

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

ROOT CAUSE

The same expert is evaluated more than once after route union.

THE FAILURE

The same expert is evaluated more than once after route union.

Unsuccessful approach: Keeping only the final contribution drops earlier route mass.

Case contract

For [token,expert,weighted_value] rows, aggregate values by (token,expert), sorted by token then expert.

Why this case matters

A deterministic toy router makes this dispatch decision independently inspectable.

1 / The failure

Exit 1

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

N = 1
observations = []
def solve(routes):
    return sorted(routes,key=lambda r:(r[0],r[1]))
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
a='a'+str(N); rows=[[0,a,N],[0,a,2*N],[1,a,N]]
check('same pair summed',solve(rows),[[0,a,3*N],[1,a,N]])
check('distinct experts',solve([[0,'b',N],[0,'a',N]]),[[0,'a',N],[0,'b',N]])
check('empty',solve([]),[])
check('negative contribution',solve([[0,'x',N],[0,'x',-1]]),[[0,'x',N-1]])
check('singleton',solve([[N,'x',N]]),[[N,'x',N]])
check('two tokens',solve([[1,'x',2],[0,'x',3]]),[[0,'x',3],[1,'x',2]])
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
same pair summed	`[[0, 'a1', 1], [0, 'a1', 2], [1, 'a1', 1]]`	`[[0, 'a1', 3], [1, 'a1', 1]]`	Failed
distinct experts	`[[0, 'a', 1], [0, 'b', 1]]`	`[[0, 'a', 1], [0, 'b', 1]]`	Passed
empty	`[]`	`[]`	Passed
negative contribution	`[[0, 'x', 1], [0, 'x', -1]]`	`[[0, 'x', 0]]`	Failed
singleton	`[[1, 'x', 1]]`	`[[1, 'x', 1]]`	Passed
two tokens	`[[0, 'x', 3], [1, 'x', 2]]`	`[[0, 'x', 3], [1, 'x', 2]]`	Passed

SHA-256 / f87f34b13e335b3fde0a1e4368b7b7c2b2c2edb2c004fff915ca6edf3f9aa7ee

2 / The unsuccessful fix

Exit 1

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

N = 1
observations = []
def solve(routes):
    d={}
    for t,e,v in routes: d[(t,e)]=v
    return [[t,e,v] for (t,e),v in sorted(d.items())]
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
a='a'+str(N); rows=[[0,a,N],[0,a,2*N],[1,a,N]]
check('same pair summed',solve(rows),[[0,a,3*N],[1,a,N]])
check('distinct experts',solve([[0,'b',N],[0,'a',N]]),[[0,'a',N],[0,'b',N]])
check('empty',solve([]),[])
check('negative contribution',solve([[0,'x',N],[0,'x',-1]]),[[0,'x',N-1]])
check('singleton',solve([[N,'x',N]]),[[N,'x',N]])
check('two tokens',solve([[1,'x',2],[0,'x',3]]),[[0,'x',3],[1,'x',2]])
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
same pair summed	`[[0, 'a1', 2], [1, 'a1', 1]]`	`[[0, 'a1', 3], [1, 'a1', 1]]`	Failed
distinct experts	`[[0, 'a', 1], [0, 'b', 1]]`	`[[0, 'a', 1], [0, 'b', 1]]`	Passed
empty	`[]`	`[]`	Passed
negative contribution	`[[0, 'x', -1]]`	`[[0, 'x', 0]]`	Failed
singleton	`[[1, 'x', 1]]`	`[[1, 'x', 1]]`	Passed
two tokens	`[[0, 'x', 3], [1, 'x', 2]]`	`[[0, 'x', 3], [1, 'x', 2]]`	Passed

SHA-256 / 378857e6684e8ddec473666aad182419b2f90eb98aed2b56af73e51a3dc925a4

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 finite toy scores and routes; the contract does not model a production model, accelerator, or learned routing 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 2026-09-30T16:48:37.646356+00:00.

Case digest / 41c56188751c7bdf28b2d6b2263ba1eb8b44b297dc47f3c5d86f106d137d6bc7