SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2212.01241 · 2022

MMBench: Benchmarking End-to-End Multi-modal DNNs and Understanding Their Hardware-Software Implications

arXiv · PDF · Open in the Atlas

Code that ran

11 of 11 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, 0 violated it and 6 have no contract verdict; 2 failures of our instrument; 0 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; 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/2212.01241.json · badge: /api/ran/2212.01241.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
xfhelen/mmbench canonical (classifier label)
classifier basis not recorded on the link
11 ran · 0 no recorded run
FunctionStatusWhere it lives
AUPRC Ran
ran, honoured the contract we drafted
xfhelen/mmbench/models/eval_scripts/performance.py
code served (permissive licence) · get_code("809e3aa9964cb3ea")
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 float, 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 sklearn.metrics

torch_seed = 0

# AUPRC expects pts, which is a list of tuples
# Each tuple is (predicted_prob, true_label)
# predicted_prob is a float (probability score)
# true_label is an int (0 or 1)

pts = [
    (0.9, 1),
    (0.2, 0),
    (0.7, 1),
    (0.3, 0),
    (0.8, 1),
    (0.1, 0),
]

ARGS = {
    "pts": pts
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, 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
create_film_layer Ran
ran on a synthesized input
xfhelen/mmbench/models/fusions/common_fusions.py
code served (permissive licence) · get_code("f2fd4bd22929f069")
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 FiLM. 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
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    "image_dim": 256,
    "text_dim": 512,
    "hidden_dim": 512,
}
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 FiLM — a FiLM output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a FiLM 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
create_film_layer_v2 Ran
ran on a synthesized input
xfhelen/mmbench/models/fusions/common_fusions.py
code served (permissive licence) · get_code("a146f39b2da4131e")
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 FiLM. 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
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    "image_dim": 4,
    "text_dim": 5,
    "audio_dim": 6,
    "hidden_dim": 8,
}
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 FiLM — a FiLM output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a FiLM 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
f1_score Ran
ran; our contract draft was wrong, not the code
xfhelen/mmbench/models/eval_scripts/performance.py
code served (permissive licence) · get_code("191c253b532edecf")
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 float, 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
import numpy as np

torch.manual_seed(0)

# Create small sample tensors for binary classification
# truth: ground truth labels
# pred: predicted labels
truth = torch.tensor([0, 1, 1, 0, 1, 1], dtype=torch.long)
pred = torch.tensor([0, 1, 0, 0, 1, 1], dtype=torch.long)

# average: the type of averaging for multi-class f1 score
average = 'weighted'

ARGS = {
    'truth': truth,
    'pred': pred,
    'average': average
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, 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
get_possible_layer_configurations Ran
ran on a synthesized input
xfhelen/mmbench/models/fusions/searchable.py
code served (permissive licence) · get_code("21907c83b0724472")
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 2, dtype family int; 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.

# Analyze the function
# get_possible_layer_configurations takes max_labels (an int or list-like)
# It recursively generates configurations by iterating through ranges
# The function treats max_labels as a sequence with len() and indexing operations
# It should be a list of integers

# Construct a simple test case
max_labels = [2, 3, 2]

ARGS = {
    "max_labels": max_labels
}
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 2, dtype family int — 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
getallparams Ran
ran, honoured the contract we drafted
xfhelen/mmbench/models/eval_scripts/complexity.py
code served (permissive licence) · get_code("f86744b3991d473e")
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 int, ndim 0, dtype family int, 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
import torch.nn as nn

torch.manual_seed(0)

# Create small neural network modules
class TinyModel1(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear1 = nn.Linear(4, 3)
        self.linear2 = nn.Linear(3, 2)
    
    def forward(self, x):
        return self.linear2(self.linear1(x))

class TinyModel2(nn.Module):
    def __init__(self):
        super().__init__()
        self.conv = nn.Conv1d(2, 3, kernel_size=2)
        self.linear = nn.Linear(3, 1)
    
    def forward(self, x):
        return self.linear(self.conv(x))

class TinyModel3(nn.Module):
    def __init__(self):
        super().__init__()
        self.embedding = nn.Embedding(10, 5)
        self.linear = nn.Linear(5, 2)
    
    def forward(self, x):
        return self.linear(self.embedding(x))

# Create instances
model1 = TinyModel1()
model2 = TinyModel2()
model3 = TinyModel3()

# Build the list of modules
module_list = [model1, model2, model3]

# Define ARGS
ARGS = {
    "li": module_list
}
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 int, ndim 0, dtype family int, 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
normalization Ran
ran; our contract draft was wrong, not the code
xfhelen/mmbench/applications/Medical-Segmentation/layers.py
pointer only (licence: MIT); the stored copy get_code serves, from guochengxiangives/AMGFormer, is pointer only (licence: NONE) · get_code("376f134d99d8ed75")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from guochengxiangives/AMGFormer, and GitHub detected no licence in that repository when it was checked. 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.

nosigmloss1d Ran
ran on a synthesized input
xfhelen/mmbench/models/objective_functions/recon.py
code served (permissive licence) · get_code("3483e2fd20eb7606")
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 Tensor, ndim 1, dtype family float, shape (2,) 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

torch.manual_seed(0)

# Create small tensors for predicted and true outputs
# a: predicted output (logits before sigmoid, values in reasonable range)
# b: true output (binary labels, values in [0, 1])
a = torch.rand(2, 4)  # batch_size=2, dim=4, values in [0, 1]
b = torch.randint(0, 2, (2, 4)).float()  # batch_size=2, dim=4, binary values

ARGS = {
    "a": a,
    "b": b,
}
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 Tensor, ndim 1, dtype family float, shape (2,) — shape (2,) 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
ptsort Ran
ran, honoured the contract we drafted
xfhelen/mmbench/models/eval_scripts/performance.py
code served (permissive licence) · get_code("c93aad3618c35efa")
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 Tensor, ndim 2, dtype family float, shape (2, 4) 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

torch.manual_seed(0)

# The function ptsort(tu) returns tu[0]
# So tu must be indexable (tuple, list, etc.) with at least one element
# The simplest valid argument is a tuple or list where the first element exists

tu = (torch.randn(2, 4), torch.randn(2, 3))

ARGS = {
    "tu": tu
}
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 Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) 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
sigmloss1d Ran
ran on a synthesized input
xfhelen/mmbench/models/objective_functions/recon.py
code served (permissive licence) · get_code("84ca4fef1e15264f")
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 Tensor, ndim 1, dtype family float, shape (2,) 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
from torch import nn

torch.manual_seed(0)

# Create small 1D tensors for predicted and true outputs
# Shape: (batch_size, features) where batch_size=2, features=4
a = torch.randn(2, 4)
b = torch.randn(2, 4)

ARGS = {
    "a": a,
    "b": b
}
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 Tensor, ndim 1, dtype family float, shape (2,) — shape (2,) 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
sigmloss1dcentercrop Ran
ran on a synthesized input
xfhelen/mmbench/models/objective_functions/recon.py
code served (permissive licence) · get_code("ad19dbde74fdba65")
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
from torch import nn

torch.manual_seed(0)

# The function sigmloss1dcentercrop takes two int parameters: adim and bdim
# Looking at the implementation:
# - adim is the predicted output size (smaller dimension)
# - bdim is the true output size (larger dimension, assumed to be > adim)
# - The function creates a loss function that crops tensors to match sizes
# - The _func inner function expects 4D tensors (batch, channels, height, width)
#   based on the indexing: b1[:, :, borderdim:bdim - borderdim, ...]
# - It flattens from start_dim=1, so it expects at least 3D tensors
# - The cropping logic assumes the spatial dimensions (dim 2 and 3) are what get compared

# Set adim smaller than bdim
adim = 4
bdim = 8

ARGS = {
    "adim": adim,
    "bdim": bdim,
}
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

Also present in other papers' code

Of 11 samples lifted from this paper's linked repositories, 11 carry a function identity and 0 do not. Of the 11 distinct functions with an identity, 5 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 0 the same month, 5 later) and 6 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
normalization 4 other papers
1 dated earlier, 3 later · 2 only as an assembly brick
2206.02425 (2022-06)
mmFormer: Multimodal Medical Transformer for Incomplete Multimodal Learning of Brain Tumor Segmentation
2303.05302 (2023-03)
M 3 AE: Multimodal Representation Learning for Brain Tumor Segmentation with Missing Modalities
2601.19349 (2026-01)
AMGFormer: Adaptive Multi-Granular Transformer for Brain Tumor Segmentation with Missing Modalities
and 1 more not named here
getallparams 2 other papers
1 dated earlier, 1 later
2107.07502 (2021-07)
MULTIBENCH: Multiscale Benchmarks for Multimodal Representation Learning
2405.17613 (2024-05)
Jointly Modeling Inter-& Intra-Modality Dependencies for Multi-modal Learning
AUPRC 1 other paper
1 later
2405.17613 (2024-05)
Jointly Modeling Inter-& Intra-Modality Dependencies for Multi-modal Learning
f1_score 1 other paper
1 later
2405.17613 (2024-05)
Jointly Modeling Inter-& Intra-Modality Dependencies for Multi-modal Learning
ptsort 1 other paper
1 later
2405.17613 (2024-05)
Jointly Modeling Inter-& Intra-Modality Dependencies for Multi-modal Learning

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 explosive growth of various types of big data and advances in AI technologies have catalyzed a new type of workloads called multi-modal DNNs. Multi-modal DNNs are capable of interpreting and reasoning about information from multiple modalities, making them more applicable to real-world AI scenarios. In recent research, multi-modal DNNs have outperformed the best uni-modal DNN in a wide range of distributed computing applications from traditional multimedia systems to emerging autonomous edge systems. However, despite their importance and superiority, very limited research attention has been devoted to understand the characteristics of multi-modal DNNs and their implications on current computing software/hardware platforms. Existing benchmarks either target uni-modal DNNs or only focus on the algorithm characteristics of multi-modal DNNs. There lacks representative benchmark suites that provide comprehensive system and architecture level analysis of multi-modal networks. To advance the understanding of these multi-modal DNN workloads and facilitate related research, we present MMBench, an open-source, end-to-end benchmark suite consisting of a set of real-world multi-modal DNN workloads with relevant performance metrics for evaluation. We then use MMBench to conduct an in-depth analysis on the characteristics of multi-modal DNNs. We demonstrate their unique characteristics of clear multi-stage execution, frequent synchronization and high heterogeneity, which distinguish them from conventional uni-modal DNNs. Finally, we conduct a case study and extend our benchmark to edge devices. We hope that our work can provide insights for future software/hardware design and optimization to underpin multi-modal DNNs on both cloud and edge computing platforms.

For agents

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

get_harvested_code_for_paper("2212.01241")
get_code_for_paper("2212.01241")
have("2212.01241")

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

curl https://syntology.ai/api/ran/2212.01241.json

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

[![Syntology run record](https://syntology.ai/api/ran/2212.01241.svg)](https://syntology.ai/paper/2212.01241)

Connect an agent — have() is free.