SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2310.08446 · ICLR · 2024

Towards Robust Multi-Modal Reasoning via Model Selection

Tao Lin, Wei Ji, Xiangyan Liu, Rongxue Li

arXiv · PDF · Open in the Atlas

Code that ran

3 of 6 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; 0 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 3 failures of our instrument; 3 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/2310.08446.json · badge: /api/ran/2310.08446.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
LINs-lab/M3 canonical (classifier label)
classifier basis: census
3 ran · 3 no recorded run
FunctionStatusWhere it lives
create_graph_from_program Ran
ran; our contract draft was wrong, not the code
LINs-lab/M3/MS-GQA/code/preprocess.py
code served (permissive licence) · get_code("f93b1f963e736c8b")
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 tuple. 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 io
import tokenize
import networkx as nx

# Define the arguments for create_graph_from_program
program = """
output1 = LOC(image=IMAGE)
output2 = VQA(image=output1)
output3 = CROP(box=output2)
output4 = COUNT(box=output3)
result = RESULT(var=output4)
""".strip()

vqa = "vqa_model"
loc = "loc_model"
node2idx = {}
type2idx = {}

ARGS = {
    "program": program,
    "vqa": vqa,
    "loc": loc,
    "node2idx": node2idx,
    "type2idx": type2idx
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
networkx — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is 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_ori_task_types Ran
ran; our contract draft was wrong, not the code
LINs-lab/M3/MS-GQA/code/run_metagl.py
code served (permissive licence) · get_code("54f795d322b7185d")
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.

import os
import json
import tempfile
import shutil

# Create a temporary directory structure for the test
temp_dir = tempfile.mkdtemp()
data_dir = os.path.join(temp_dir, 'data')
os.makedirs(data_dir, exist_ok=True)

# Create the test data file
test_data = {
    "q1": {
        "imageId": "img_001",
        "question": "What is this?",
        "types": ["object"]
    },
    "q2": {
        "imageId": "img_002",
        "question": "Where is it?",
        "types": ["spatial"]
    },
    "q3": {
        "imageId": "img_003",
        "question": "How many?",
        "types": ["count"]
    }
}

file_path = os.path.join(data_dir, 'testdev_balanced_questions.json')
with open(file_path, 'w') as f:
    json.dump(test_data, f)

# Change to the temp directory so relative path works
original_cwd = os.getcwd()
os.chdir(temp_dir)

# Create merged_list that will be updated by the function
merged_list = [
    {
        "imageId": "img_001",
        "question": "What is this?",
        "types": []
    },
    {
        "imageId": "img_002",
        "question": "Where is it?",
        "types": []
    },
    {
        "imageId": "img_004",
        "question": "Unknown question?",
        "types": []
    }
]

ARGS = {
    "merged_list": merged_list
}
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
parse_step Ran
ran; our contract draft was wrong, not the code
LINs-lab/M3/MS-GQA/code/preprocess.py
code served (permissive licence) · get_code("6789f8911c900d76")
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 dict. 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 io
import tokenize

torch_imported = False
try:
    import torch
    torch_imported = True
except:
    pass

# Test case 1: Simple assignment with arguments
step_str_full = "output1 = step_name(arg1=value1, arg2=value2)"

# Test case 2: Simple assignment without arguments (partial=True)
step_str_partial = "output2 = another_step()"

ARGS = {
    "step_str": step_str_full,
    "partial": False
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
build_dataset Not yet run
unverified (harvested, no recorded run)
LINs-lab/M3/MS-GQA/code/run_metagl.py
code served (permissive licence) · get_code("d8d9625bb4bcbddf")
How it ran

no invocation recorded.

evaluate_model Not yet run
unverified (harvested, no recorded run)
LINs-lab/M3/MS-GQA/code/run_m3.py
code served (permissive licence) · get_code("a1cb3bb3f6e48673")
How it ran

no invocation recorded.

evaluate_model Not yet run
unverified (harvested, no recorded run)
LINs-lab/M3/MS-GQA/code/run_ncf++.py
code served (permissive licence) · get_code("80cf4f2a02aa0335")
How it ran

no invocation recorded.

Also present in other papers' code

Of 6 samples lifted from this paper's linked repositories, 6 carry a function identity and 0 do not. Of the 6 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 1 later) and 5 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.

FunctionAlso present inEarliest-datedOthers named
parse_step 2 other papers
1 dated earlier, 1 later
2211.11559 (2022-11)
Visual Programming: Compositional visual reasoning without training
2311.04901 (2023-11)
GENOME: GenerativE Neuro-symbOlic visual reasoning by growing and reusing ModulEs

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

The reasoning capabilities of LLM (Large Language Model) are widely acknowledged in recent research, inspiring studies on tool learning and autonomous agents. LLM serves as the "brain" of the agent, orchestrating multiple tools for collaborative multi-step task solving. Unlike methods invoking tools like calculators or weather APIs for straightforward tasks, multi-modal agents excel by integrating diverse AI models for complex challenges. However, current multi-modal agents neglect the significance of model selection: they primarily focus on the planning and execution phases, and will mainly invoke predefined task-specific models for each subtask, making the execution fragile. Meanwhile, other traditional model selection methods are either incompatible with or suboptimal for the multi-modal agent scenarios, due to ignorance of dependencies among subtasks arising by multi-step reasoning. To this end, we identify the key challenges therein and propose the M 3 framework as a plug-in with negligible runtime overhead at test-time. This framework improves model selection and bolsters the robustness of multi-modal agents in multi-step reasoning. In the absence of suitable benchmarks, we create MS-GQA, a new dataset specifically designed to investigate the model selection challenge in multi-modal agents. Our experiments reveal that our framework enables dynamic model selection, considering both user inputs and subtask dependencies, thereby robustifying the overall reasoning process. Our code and benchmark: https://github.com/LINs-lab/M3.

For agents

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

get_harvested_code_for_paper("2310.08446")
get_code_for_paper("2310.08446")
have("2310.08446")

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

curl https://syntology.ai/api/ran/2310.08446.json

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

[![Syntology run record](https://syntology.ai/api/ran/2310.08446.svg)](https://syntology.ai/paper/2310.08446)

Connect an agent — have() is free.