SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2108.11898 · 2021

Supervised Compression for Resource-Constrained Edge Computing Systems

arXiv · PDF · Open in the Atlas

Code that ran

11 of 17 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 10 have no contract verdict; 1 failure of our instrument; 6 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/2108.11898.json · badge: /api/ran/2108.11898.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
yoshitomo-matsubara/supervised-compression canonical (classifier label)
classifier basis: census
2 ran · 2 no recorded run
yoshitomo-matsubara/sc2-benchmark community (classifier label)
classifier basis not recorded on the link
8 ran · 4 no recorded run
copy not recorded community
not in the paper's declared repositories
1 ran · 0 no recorded run
FunctionStatusWhere it lives
check_if_analyzable Ran
ran on a synthesized input
yoshitomo-matsubara/sc2-benchmark/sc2bench/analysis.py
code served (permissive licence) · get_code("cd97b267e4c63c03")
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
from torch import nn

torch.manual_seed(0)

class SimpleModule(nn.Module):
    """A simple module for testing."""
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.linear(x)

class AnalyzableModuleExample(nn.Module):
    """An example of an AnalyzableModule."""
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.linear(x)

# Create instances for testing
simple_module = SimpleModule()
analyzable_module = AnalyzableModuleExample()

ARGS = {
    "module": simple_module
}
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
compute_accuracy Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
this paper's copy was not recorded; identical code first harvested from yoshitomo-matsubara/torchdistill
pointer only · get_code("a427a6169de8f607")
How it ran

The body of this sample is withheld: this paper's copy of the code was not recorded (its link names no repository or source path), so no licence was read for it. 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.

register_analysis_class Ran
ran on a synthesized input
yoshitomo-matsubara/sc2-benchmark/sc2bench/analysis.py
code served (permissive licence) · get_code("fbb9004c52a74063")
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 type. 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

# Clear the dict to start fresh
ANALYZER_CLASS_DICT = dict()

def register_analysis_class(cls):
    """
    Registers an analyzer class.

    :param cls: analyzer class to be registered
    :type cls: class
    :return: registered analyzer class
    :rtype: class
    """
    ANALYZER_CLASS_DICT[cls.__name__] = cls
    return cls


# Create a minimal analyzer class to register
class SimpleAnalyzer:
    """A simple stand-in analyzer class for testing."""
    def __init__(self):
        self.name = "SimpleAnalyzer"
    
    def analyze(self, data):
        return data


# Define the arguments for the function
ARGS = {
    "cls": SimpleAnalyzer
}
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 type — a type output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a type 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
register_compressai_model Ran
ran on a synthesized input
yoshitomo-matsubara/sc2-benchmark/sc2bench/models/registry.py
code served (permissive licence) · get_code("39ed763e55dbf0e9")
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 type. 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

# Clear the registry for clean state
COMPRESSAI_DICT = dict()

# Define a simple compression model class
class SimpleCompressionModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 2)
    
    def forward(self, x):
        return self.linear(x)

# Define a simple function to build a compression model
def build_compression_model():
    return SimpleCompressionModel()

# Test with class
test_class = SimpleCompressionModel

# Test with function
test_func = build_compression_model

ARGS = {
    "cls_or_func": test_class
}
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 type — a type output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a type 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
register_compression_model_class Ran
ran on a synthesized input
yoshitomo-matsubara/supervised-compression/compression/registry.py
code served (permissive licence) · get_code("0d5cd1d5efbacf46")
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 type. 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)

# The function register_compression_model_class takes one parameter: cls
# It uses cls.__name__ to get the class name and stores the class in a dict
# So we need to pass a class (any class will do)

class DummyCompressionModel(nn.Module):
    """A minimal compression model class for testing"""
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.linear(x)

ARGS = {
    'cls': DummyCompressionModel
}
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 type — a type output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a type 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
register_compression_model_class Ran
ran on a synthesized input
yoshitomo-matsubara/sc2-benchmark/sc2bench/models/registry.py
code served (permissive licence) · get_code("733180f8f32962be")
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 type. 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

COMPRESSION_MODEL_CLASS_DICT = dict()

def register_compression_model_class(cls):
    """
    Registers a compression model class.

    :param cls: compression model to be registered
    :type cls: class
    :return: registered compression model class
    :rtype: class
    """
    COMPRESSION_MODEL_CLASS_DICT[cls.__name__] = cls
    return cls


# Create a simple compression model class to register
class SimpleCompressionModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 2)
    
    def forward(self, x):
        return self.linear(x)


ARGS = {
    "cls": SimpleCompressionModel
}
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 type — a type output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a type 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
register_compression_model_func Ran
ran on a synthesized input
yoshitomo-matsubara/supervised-compression/compression/registry.py
code served (permissive licence) · get_code("32164b3ce42adca0")
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
import torch.nn as nn

torch.manual_seed(0)

# The function register_compression_model_func takes a single parameter: func
# It expects func to be a callable (function or callable object)
# The function accesses func.__name__ and stores it in a dict
# So we need to provide a function with a __name__ attribute

def sample_compression_model():
    """A simple stand-in compression model function"""
    pass

ARGS = {
    "func": sample_compression_model
}
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
register_compression_model_func Ran
ran on a synthesized input
yoshitomo-matsubara/sc2-benchmark/sc2bench/models/registry.py
code served (permissive licence) · get_code("46791505e5924b54")
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
import torch.nn as nn

torch.manual_seed(0)

# Create a simple compression model function to register
def simple_compression_model():
    """A simple stand-in compression model builder function."""
    return nn.Linear(4, 3)

# The function register_compression_model_func takes one parameter: func
# It expects func to be a callable that builds a compression model
# The function adds it to COMPRESSION_MODEL_FUNC_DICT with func.__name__ as key

ARGS = {
    "func": simple_compression_model
}
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
register_layer_class Ran
ran on a synthesized input
yoshitomo-matsubara/sc2-benchmark/sc2bench/models/layer.py
code served (permissive licence) · get_code("ead935a341e1f176")
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 type. 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 a minimal layer class for testing
class SimpleLayer(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.linear(x)

# The function only needs the cls parameter
# It accesses cls.__name__ and stores it in the dict
ARGS = {
    "cls": SimpleLayer
}
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 type — a type output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a type 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
register_layer_func Ran
ran on a synthesized input
yoshitomo-matsubara/sc2-benchmark/sc2bench/models/layer.py
code served (permissive licence) · get_code("92d43e7ee8bfedda")
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
import torch.nn as nn

torch.manual_seed(0)

LAYER_FUNC_DICT = dict()

def register_layer_func(func):
    """
    Registers a function to build a layer module.

    :param func: function to build a layer module
    :type func: typing.Callable
    :return: registered function
    :rtype: typing.Callable
    """
    LAYER_FUNC_DICT[func.__name__] = func
    return func

def sample_layer_builder():
    """A simple function that builds a layer module."""
    return nn.Linear(4, 3)

ARGS = {
    'func': sample_layer_builder
}
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
register_wrapper_class Ran
ran on a synthesized input
yoshitomo-matsubara/sc2-benchmark/sc2bench/models/wrapper.py
code served (permissive licence) · get_code("29dce4ec3c15a305")
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 type. 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

class DummyModelWrapper(nn.Module):
    """A simple wrapper class for testing registration."""
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.linear(x)

ARGS = {
    'cls': DummyModelWrapper
}
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 type — a type output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a type 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_faster_rcnn_fpn Not yet run
unverified (harvested, no recorded run)
yoshitomo-matsubara/sc2-benchmark/sc2bench/models/detection/rcnn.py
code served (permissive licence) · get_code("032512a773a6431a")
How it ran

no invocation recorded.

get_analyzer Not yet run
unverified (harvested, no recorded run)
yoshitomo-matsubara/sc2-benchmark/sc2bench/analysis.py
code served (permissive licence) · get_code("2153756735456ef5")
How it ran

no invocation recorded.

get_compression_model Not yet run
unverified (harvested, no recorded run)
yoshitomo-matsubara/supervised-compression/compression/registry.py
code served (permissive licence) · get_code("13699b49bbd77789")
How it ran

no invocation recorded.

get_detection_model Not yet run
unverified (harvested, no recorded run)
yoshitomo-matsubara/sc2-benchmark/sc2bench/models/detection/registry.py
code served (permissive licence) · get_code("edb9af2bfe3c3d07")
How it ran

no invocation recorded.

load_ckpt Not yet run
unverified (harvested, no recorded run)
yoshitomo-matsubara/supervised-compression/ckpt_editor.py
code served (permissive licence) · get_code("5bf69d1c652d53f2")
How it ran

no invocation recorded.

wrap_model Not yet run
unverified (harvested, no recorded run)
yoshitomo-matsubara/sc2-benchmark/sc2bench/models/wrapper.py
code served (permissive licence) · get_code("09b688d30bb685dc")
How it ran

no invocation recorded.

Also present in other papers' code

Of 17 samples lifted from this paper's linked repositories, 17 carry a function identity and 0 do not. Of the 15 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, 0 later) and 14 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
compute_accuracy 5 other papers
5 dated earlier
1412.6550 (2014-12)
FitNets: Hints for Thin Deep Nets
1503.02531 (2015-03)
Distilling the Knowledge in a Neural Network
1904.05068 (2019-04)
Relational Knowledge Distillation
and 2 more not named here

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

There has been much interest in deploying deep learning algorithms on low-powered devices, including smartphones, drones, and medical sensors. However, full-scale deep neural networks are often too resource-intensive in terms of energy and storage. As a result, the bulk part of the machine learning operation is therefore often carried out on an edge server, where the data is compressed and transmitted. However, compressing data (such as images) leads to transmitting information irrelevant to the supervised task. Another popular approach is to split the deep network between the device and the server while compressing intermediate features. To date, however, such split computing strategies have barely outperformed the aforementioned naive data compression baselines due to their inefficient approaches to feature compression. This paper adopts ideas from knowledge distillation and neural image compression to compress intermediate feature representations more efficiently. Our supervised compression approach uses a teacher model and a student model with a stochastic bottleneck and learnable prior for entropy coding (Entropic Student). We compare our approach to various neural image and feature compression baselines in three vision tasks and found that it achieves better supervised rate-distortion performance while maintaining smaller end-to-end latency. We furthermore show that the learned feature representations can be tuned to serve multiple downstream tasks.

For agents

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

get_harvested_code_for_paper("2108.11898")
get_code_for_paper("2108.11898")
have("2108.11898")

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

curl https://syntology.ai/api/ran/2108.11898.json

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

[![Syntology run record](https://syntology.ai/api/ran/2108.11898.svg)](https://syntology.ai/paper/2108.11898)

Connect an agent — have() is free.