SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2602.13792 · 2026

StackingNet: Collective Inference across Independent AI Foundation Models

Zhigang Zeng, Siyang Li, Chenhao Liu, † Id Dongrui Wu, Lieyun Ding

arXiv · PDF

Code that ran

12 of 16 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 3 honoured the contract we drafted, 1 violated it and 0 have no contract verdict; 8 failures of our instrument; 4 no recorded run), measured 2026-11-22; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2602.13792.json · badge: /api/ran/2602.13792.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
pymupdf/PyMuPDF canonical (classifier label)
classifier basis: census
8 ran · 3 no recorded run
openreview/openreview-py canonical (classifier label)
classifier basis: census
4 ran · 1 no recorded run
FunctionStatusWhere it lives
chars_in_rect Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
pymupdf/PyMuPDF/src/table.py
pointer only (licence: AGPL-3.0) · get_code("7dd9969debc51bf8")
How it ran

The body of this sample is withheld: AGPL-3.0, as GitHub detects it on this paper's repository, pymupdf/PyMuPDF, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

count_alpha Ran
ran, honoured the contract we drafted
pymupdf/PyMuPDF/src/_table_headers.py
pointer only (licence: AGPL-3.0) · get_code("a02e9552a4277294")
How it ran

The body of this sample is withheld: AGPL-3.0, as GitHub detects it on this paper's repository, pymupdf/PyMuPDF, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

count_digit Ran
ran, honoured the contract we drafted
pymupdf/PyMuPDF/src/_table_headers.py
pointer only (licence: AGPL-3.0) · get_code("8bd5338eb9d52857")
How it ran

The body of this sample is withheld: AGPL-3.0, as GitHub detects it on this paper's repository, pymupdf/PyMuPDF, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

decision_to_venue Ran
ran; our contract draft was wrong, not the code
openreview/openreview-py/openreview/tools.py
code served (permissive licence) · get_code("fbe9624c155dee52")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import re

def is_accept_decision(decision, accept_options=None):
    """
    Checks if decision is an accept decision

    :param decision: paper decision (i.e., Accept, Reject)
    :type decision: string
    :param accept_options: accept decisions (i.e., [ Accept (Best Paper), Invite to Archive ])
    :type accept_options: list
    """
    if accept_options and decision in accept_options or (not accept_options and 'Accept' in decision):
        return True
    return False

def decision_to_venue(venue_id, decision_option, accept_options=None):
    """
    Returns the venue for a submission based on its decision

    :param venue_id: venue's short name (i.e., ICLR 2022)
    :type venue_id: string
    :param decision_option: paper decision (i.e., Accept, Reject)
    :type decision_option: string
    :param accept_options: accept decisions (i.e., [ Accept (Best Paper), Invite to Archive ])
    :type accept_options: list
    """
    venue = venue_id
    if is_accept_decision(decision_option, accept_options):
        decision = decision_option.replace('Accept', '') if 'Accept' in decision_option else decision_option
        decision = re.sub('[()\\W]+', '', decision)
        if decision:
            venue += ' ' + decision.strip()
    else:
        venue = f'Submitted to {venue}'
    return venue

ARGS = {
    'venue_id': 'ICLR 2022',
    'decision_option': 'Accept (Best Paper)',
    'accept_options': ['Accept', 'Accept (Best Paper)', 'Invite to Archive']
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_list Ran
ran, honoured the contract we drafted
pymupdf/PyMuPDF/src_classic/__main__.py
pointer only (licence: AGPL-3.0) · get_code("9b0844c695b3f09d")
How it ran

The body of this sample is withheld: AGPL-3.0, as GitHub detects it on this paper's repository, pymupdf/PyMuPDF, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_resubmissions Ran
ran; our contract draft was wrong, not the code
openreview/openreview-py/openreview/arr/helpers.py
code served (permissive licence) · get_code("74c52f39a08a4de7")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

class MockSubmission:
    def __init__(self, content):
        self.content = content

# Create test submissions with various content structures
submission1 = MockSubmission({
    'previous_url': {'value': ['https://example.com/page1']},
    'other_field': 'data'
})

submission2 = MockSubmission({
    'previous_url': {'value': []},
    'other_field': 'data'
})

submission3 = MockSubmission({
    'previous_url': {'value': ['https://example.com/page2', 'https://example.com/page3']},
    'other_field': 'data'
})

submission4 = MockSubmission({
    'different_field': 'data'
})

submission5 = MockSubmission({
    'previous_url': {'no_value_key': 'something'},
    'other_field': 'data'
})

submissions = [submission1, submission2, submission3, submission4, submission5]
previous_url_field = 'previous_url'

ARGS = {
    'submissions': submissions,
    'previous_url_field': previous_url_field
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
intersects_words_h Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
pymupdf/PyMuPDF/src/table.py
pointer only (licence: AGPL-3.0) · get_code("2a5dbed3ca23ac18")
How it ran

The body of this sample is withheld: AGPL-3.0, as GitHub detects it on this paper's repository, pymupdf/PyMuPDF, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

is_accept_decision Ran
ran, violated the contract we drafted
openreview/openreview-py/openreview/tools.py
code served (permissive licence) · get_code("66abbbd367df035b")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

ARGS = {
    "decision": "Accept",
    "accept_options": ["Accept", "Accept (Best Paper)", "Invite to Archive"]
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
mycenter Ran
ran; our contract draft was wrong, not the code
pymupdf/PyMuPDF/src/__main__.py
pointer only (licence: AGPL-3.0) · get_code("1da4e7527eb4ea1b")
How it ran

The body of this sample is withheld: AGPL-3.0, as GitHub detects it on this paper's repository, pymupdf/PyMuPDF, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

rect_in_rect Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
pymupdf/PyMuPDF/src/table.py
pointer only (licence: AGPL-3.0) · get_code("4994b1b9c7f2487d")
How it ran

The body of this sample is withheld: AGPL-3.0, as GitHub detects it on this paper's repository, pymupdf/PyMuPDF, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

run_once Ran
ran; our contract draft was wrong, not the code
openreview/openreview-py/openreview/tools.py
code served (permissive licence) · get_code("28d56af3f40fa05d")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned function. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

def sample_function(x):
    """A simple function to decorate"""
    return x * 2 + 1

ARGS = {
    'f': sample_function
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
tokens Ran
ran; our contract draft was wrong, not the code
pymupdf/PyMuPDF/src/_table_headers.py
pointer only (licence: AGPL-3.0) · get_code("c3b568421a889173")
How it ran

The body of this sample is withheld: AGPL-3.0, as GitHub detects it on this paper's repository, pymupdf/PyMuPDF, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_conference Not yet run
unverified (harvested, no recorded run)
openreview/openreview-py/openreview/helpers.py
code served (permissive licence) · get_code("84d627efcd402fae")
How it ran

no invocation recorded.

insert_image Not yet run
unverified (harvested, no recorded run)
pymupdf/PyMuPDF/src_classic/utils.py
pointer only (licence: AGPL-3.0) · get_code("b39153247c44a759")
How it ran

The body of this sample is withheld: AGPL-3.0, as GitHub detects it on this paper's repository, pymupdf/PyMuPDF, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

open_file Not yet run
unverified (harvested, no recorded run)
pymupdf/PyMuPDF/src/__main__.py
pointer only (licence: AGPL-3.0) · get_code("90fe26e196474695")
How it ran

The body of this sample is withheld: AGPL-3.0, as GitHub detects it on this paper's repository, pymupdf/PyMuPDF, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

recoverpix Not yet run
unverified (harvested, no recorded run)
pymupdf/PyMuPDF/src/__main__.py
pointer only (licence: AGPL-3.0) · get_code("1836674318f018ee")
How it ran

The body of this sample is withheld: AGPL-3.0, as GitHub detects it on this paper's repository, pymupdf/PyMuPDF, is not in the permissive set Syntology serves inline. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

Also present in other papers' code

Of 16 samples lifted from this paper's linked repositories, 16 carry a function identity and 0 do not. Of the 16 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 16 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Artificial intelligence built on large foundation models has transformed language understanding, computer vision, and reasoning, yet these systems remain isolated and cannot readily share their capabilities. Coordinating the complementary strengths of independently developed, black-box foundation models is essential for trustworthy intelligent systems, yet no established method exists. Here we show that such coordination can be achieved through a meta-ensemble framework termed StackingNet, which aggregates the output predictions of independent models at inference. StackingNet improves accuracy, reduces individual-model error and group-wise disparities, ranks model reliability, and identifies or prunes models that degrade performance, all without access to internal parameters or training data. Across language comprehension, visual attribute estimation, and academic paper rating, it consistently outperforms individual models and classic ensembles, with gains that persist when the base models are uniformly strong. These gains stem from variance reduction and consensus alignment among independent models rather than from any emergent group cognition, and they widen as the model pool grows more diverse. By turning model diversity from a source of inconsistency into a resource for cooperation, Stack-ingNet offers a practical path toward coordinated artificial intelligence, where progress emerges not only from larger single models but from principled cooperation among many specialized ones.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2602.13792")
get_code_for_paper("2602.13792")
have("2602.13792")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2602.13792.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2602.13792.svg)](https://syntology.ai/paper/2602.13792)

Connect an agent — have() is free.