SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2312.02829 · NeurIPS · 2023

MIMONets: Multiple-Input-Multiple-Output Neural Networks Exploiting Computation in Superposition

Nicolas Menet, Michael Hersche, Geethan Karunaratne, Luca Benini, Abu Sebastian

arXiv · PDF · Open in the Atlas

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 (8 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 8 have no contract verdict; 4 failures of our instrument; 4 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/2312.02829.json · badge: /api/ran/2312.02829.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
ibm/multiple-input-multiple-output-nets canonical (classifier label)
classifier basis: classifier_2025
8 ran · 3 no recorded run
IBM/multiple-input-multiple-output-nets canonical (classifier label)
classifier basis: census
4 ran · 1 no recorded run
FunctionStatusWhere it lives
AdjustedISOBlock Ran
ran on a synthesized input · constructed an object; no result was computed
ibm/multiple-input-multiple-output-nets/MIMOConv/src/models/superwideisonet.py
code served (permissive licence) · get_code("669d4ff1599a17f0")
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 an instance of its own class, AdjustedISOBlock: the run constructed the object and did not call it, so no result was computed. 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)

# AdjustedISOBlock parameters:
# - inplanes: int
# - planes: int
# - stride: int (default 1)
# - downsample: Optional[nn.Module] (default None)
# - norm_layer: Optional[nn.Module] (default None)
# - relu_parameter: float (default 0.5)
# - skip_init: bool (default False)

# Create a minimal downsample module that takes input with inplanes channels
# and outputs planes channels. The forward method is called in the block's forward.
class MinimalDownsample(nn.Module):
    def __init__(self, inplanes: int, planes: int, stride: int):
        super().__init__()
        self.conv = nn.Conv2d(inplanes, planes, kernel_size=1, stride=stride, bias=False)
    
    def forward(self, x):
        return self.conv(x)

# Define arguments for AdjustedISOBlock
inplanes_val = 4
planes_val = 8
stride_val = 2
downsample_val = MinimalDownsample(inplanes_val, planes_val, stride_val)
norm_layer_val = nn.BatchNorm2d
relu_parameter_val = 0.5
skip_init_val = False

ARGS = {
    "inplanes": inplanes_val,
    "planes": planes_val,
    "stride": stride_val,
    "downsample": downsample_val,
    "norm_layer": norm_layer_val,
    "relu_parameter": relu_parameter_val,
    "skip_init": skip_init_val,
}
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 AdjustedISOBlock — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a AdjustedISOBlock 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
BasicBlock Ran
ran on a synthesized input · constructed an object; no result was computed
ibm/multiple-input-multiple-output-nets/MIMOConv/src/models/superwideisonet.py
code served (permissive licence) · get_code("efa72260d78717d1")
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 an instance of its own class, BasicBlock: the run constructed the object and did not call it, so no result was computed. 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)

# Create a minimal downsample module that the forward pass expects
downsample_module = nn.Sequential(
    nn.Conv2d(4, 3, kernel_size=1, stride=2, bias=False),
    nn.BatchNorm2d(3)
)

# Define arguments for BasicBlock
ARGS = {
    "inplanes": 4,
    "planes": 3,
    "stride": 2,
    "downsample": downsample_module,
    "norm_layer": nn.BatchNorm2d,
    "relu_parameter": None,
    "skip_init": 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 BasicBlock — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a BasicBlock 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
BasicISOBlock Ran
ran on a synthesized input · constructed an object; no result was computed
ibm/multiple-input-multiple-output-nets/MIMOConv/src/models/superwideisonet.py
code served (permissive licence) · get_code("7de9f896931b5aae")
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 an instance of its own class, BasicISOBlock: the run constructed the object and did not call it, so no result was computed. 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
from torch import Tensor

torch.manual_seed(0)

# Create a small downsample module (1x1 conv) for when stride > 1 or channels change
downsample_module = nn.Sequential(
    nn.Conv2d(4, 8, kernel_size=1, stride=2, bias=False),
    nn.BatchNorm2d(8)
)

# Create a custom norm layer
norm_layer = nn.BatchNorm2d

ARGS = {
    "inplanes": 4,
    "planes": 8,
    "stride": 2,
    "downsample": downsample_module,
    "norm_layer": norm_layer,
    "relu_parameter": -1,
    "skip_init": 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 BasicISOBlock — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a BasicISOBlock 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
SReLU Ran
ran on a synthesized input · constructed an object; no result was computed
ibm/multiple-input-multiple-output-nets/MIMOConv/src/models/superwideisonet.py
code served (permissive licence) · get_code("3105bc27785bc6fa")
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 an instance of its own class, SReLU: the run constructed the object and did not call it, so no result was computed. 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
from torch import Tensor

torch.manual_seed(0)

# Construct ARGS dict with all parameters for SReLU.__init__
ARGS = {
    'nc': 3,
    'relu_parameter': -0.5
}
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 SReLU — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a SReLU 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
Superposition Ran
ran on a synthesized input · constructed an object; no result was computed
ibm/multiple-input-multiple-output-nets/MIMOConv/src/models/superwideisonet.py
code served (permissive licence) · get_code("59141695e5f08541")
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 an instance of its own class, Superposition: the run constructed the object and did not call it, so no result was computed. 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)

# Define ARGS dict with all parameters for Superposition.__init__
ARGS = {
    'num_img_sup_cap': 2,
    'binding_type': 'HRR',
    'channels': 4,
    'fully_connected_features': 8,
    'trainable_keys': True
}
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 Superposition — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a Superposition 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
append_cls Ran
ran; our contract draft was wrong, not the code
IBM/multiple-input-multiple-output-nets/MIMOFormer/code/model_wrapper.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from sc782/SBM-Transformer, is pointer only (licence: NONE) · get_code("defcc1e5e9bf48b2")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from sc782/SBM-Transformer, 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.

mixup_criterion Ran
ran; our contract draft was wrong, not the code
IBM/multiple-input-multiple-output-nets/MIMOConv/src/mixup.py
code served (permissive licence) · get_code("9b48dc4a035b2412")
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 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 torch.nn as nn

torch.manual_seed(0)

# Create a simple criterion (loss function)
criterion = nn.CrossEntropyLoss()

# Create predicted logits: batch_size=2, num_classes=4
pred = torch.randn(2, 4)

# Create ground truth labels for first input: batch_size=2
y_a = torch.tensor([0, 1])

# Create ground truth labels for second input: batch_size=2
y_b = torch.tensor([2, 3])

# Create lambda weight (between 0 and 1)
lam = 0.6

ARGS = {
    "criterion": criterion,
    "pred": pred,
    "y_a": y_a,
    "y_b": y_b,
    "lam": lam
}
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 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
mixup_data Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
IBM/multiple-input-multiple-output-nets/MIMOConv/src/mixup.py
code served (permissive licence) · get_code("112a7683003fa463")
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. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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 numpy as np
import torch

torch.manual_seed(0)

# Create small tensors for input data and labels
x = torch.randn(2, 4)  # batch_size=2, features=4
y = torch.tensor([0, 1])  # labels for batch_size=2

# Define ARGS dict with all parameters
ARGS = {
    'x': x,
    'y': y,
    'alpha': 1.0,
    'use_cuda': False,  # CPU only
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — 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
noOrthoRegularization Ran
ran on a synthesized input · constructed an object; no result was computed
ibm/multiple-input-multiple-output-nets/MIMOConv/src/models/superwideisonet.py
code served (permissive licence) · get_code("544ec69df8400808")
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 an instance of its own class, noOrthoRegularization: the run constructed the object and did not call it, so no result was computed. 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 = {}
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 noOrthoRegularization — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a noOrthoRegularization 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
noOrthoRegularizationConv1d Ran
ran on a synthesized input · constructed an object; no result was computed
ibm/multiple-input-multiple-output-nets/MIMOConv/src/models/superwideisonet.py
code served (permissive licence) · get_code("57b78497fa0083bf")
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 an instance of its own class, noOrthoRegularizationConv1d: the run constructed the object and did not call it, so no result was computed. 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)

# noOrthoRegularizationConv1d inherits from nn.Conv1d
# nn.Conv1d signature:
# Conv1d(in_channels, out_channels, kernel_size, stride=1, padding=0, dilation=1, groups=1, bias=True, padding_mode='zeros', device=None, dtype=None)

ARGS = {
    'in_channels': 4,
    'out_channels': 3,
    'kernel_size': 5,
    'stride': 1,
    'padding': 0,
    'dilation': 1,
    'groups': 1,
    'bias': True,
    'padding_mode': 'zeros',
    'device': None,
    'dtype': None,
}
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 noOrthoRegularizationConv1d — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a noOrthoRegularizationConv1d 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
noOrthoRegularizationConv2d Ran
ran on a synthesized input · constructed an object; no result was computed
ibm/multiple-input-multiple-output-nets/MIMOConv/src/models/superwideisonet.py
code served (permissive licence) · get_code("aa2e68450b551710")
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 an instance of its own class, noOrthoRegularizationConv2d: the run constructed the object and did not call it, so no result was computed. 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 = {
    'in_channels': 3,
    'out_channels': 16,
    'kernel_size': 3,
    'stride': 1,
    'padding': 1,
    'dilation': 1,
    'groups': 1,
    'bias': True,
    'padding_mode': 'zeros',
}
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 noOrthoRegularizationConv2d — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a noOrthoRegularizationConv2d 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
pooling Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
IBM/multiple-input-multiple-output-nets/MIMOFormer/code/model_wrapper.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from sc782/SBM-Transformer, is pointer only (licence: NONE) · get_code("9dbe45c9859b04e5")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from sc782/SBM-Transformer, 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.

SuperWideISONet Not yet run
unverified (harvested, no recorded run)
ibm/multiple-input-multiple-output-nets/MIMOConv/src/models/superwideisonet.py
code served (permissive licence) · get_code("425195ab71f03e63")
How it ran

no invocation recorded.

SuperWideResnet Not yet run
unverified (harvested, no recorded run)
ibm/multiple-input-multiple-output-nets/MIMOConv/src/models/superwideisonet.py
code served (permissive licence) · get_code("e0f94d0375d9e207")
How it ran

no invocation recorded.

get_train_eval_test_sets Not yet run
unverified (harvested, no recorded run)
IBM/multiple-input-multiple-output-nets/MIMOConv/src/datasets.py
code served (permissive licence) · get_code("88df50dd66680264")
How it ran

no invocation recorded.

train_epoch Not yet run
unverified (harvested, no recorded run)
ibm/multiple-input-multiple-output-nets/MIMOConv/src/train_superposition.py
code served (permissive licence) · get_code("dd579f5afe72bd6a")
How it ran

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, 4 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 0 the same month, 4 later) and 12 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
mixup_criterion 14 other papers
10 dated earlier, 4 later
1706.06083 (2017-06)
Towards Deep Learning Models Resistant to Adversarial Attacks
1905.02175 (2019-05)
Adversarial Examples Are Not Bugs, They Are Features
1906.06784 (2019-06)
Interpolated Adversarial Training: Achieving Robust Neural Networks without Sacrificing Too Much Accuracy
and 11 more not named here
mixup_data 12 other papers
9 dated earlier, 3 later
1908.09699 (2019-08)
Gated Convolutional Networks with Hybrid Connectivity for Image Classification
2102.13280 (2021-02)
MixSearch: Searching for Domain Generalized Medical Image Segmentation Architectures
2104.00466 (2021-04)
Improving Calibration for Long-Tailed Recognition
and 9 more not named here
append_cls 1 other paper
1 later
2607.21752 (2026-07)
Parameter-free Adaptive Sparse Attention via Compression-Based Content Selection
pooling 1 other paper
1 later
2607.21752 (2026-07)
Parameter-free Adaptive Sparse Attention via Compression-Based Content Selection

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

With the advent of deep learning, progressively larger neural networks have been designed to solve complex tasks. We take advantage of these capacity-rich models to lower the cost of inference by exploiting computation in superposition. To reduce the computational burden per input, we propose Multiple-Input-Multiple-Output Neural Networks (MIMONets) capable of handling many inputs at once. MIMONets augment various deep neural network architectures with variable binding mechanisms to represent an arbitrary number of inputs in a compositional data structure via fixed-width distributed representations. Accordingly, MIMONets adapt nonlinear neural transformations to process the data structure holistically, leading to a speedup nearly proportional to the number of superposed input items in the data structure. After processing in superposition, an unbinding mechanism recovers each transformed input of interest. MIMONets also provide a dynamic trade-off between accuracy and throughput by an instantaneous on-demand switching between a set of accuracy-throughput operating points, yet within a single set of fixed parameters. We apply the concept of MIMONets to both CNN and Transformer architectures resulting in MIMOConv and MIMOFormer, respectively. Empirical evaluations show that MIMOConv achieves ≈ 2 -4× speedup at an accuracy delta within [+0.68, -3.18]% compared to WideResNet CNNs on CI-FAR10 and CIFAR100. Similarly, MIMOFormer can handle 2-4 inputs at once while maintaining a high average accuracy within a [-1.07, -3.43]% delta on the long range arena benchmark. Finally, we provide mathematical bounds on the interference between superposition channels in MIMOFormer. Our code is available at https://github.com/IBM/multiple-input-multiple-output-nets.

For agents

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

get_harvested_code_for_paper("2312.02829")
get_code_for_paper("2312.02829")
have("2312.02829")

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

curl https://syntology.ai/api/ran/2312.02829.json

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

[![Syntology run record](https://syntology.ai/api/ran/2312.02829.svg)](https://syntology.ai/paper/2312.02829)

Connect an agent — have() is free.