SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2311.09058 · NeurIPS · 2024

Improving Deep Learning Optimization through Constrained Parameter Regularization

Frank Hutter, Jörg Franke, Gregor Koehler, Michael Hefenbrock

arXiv · PDF · Open in the Atlas

Code that ran

4 of 13 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; 4 failures of our instrument; 9 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/2311.09058.json · badge: /api/ran/2311.09058.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
automl/cpr canonical (classifier label)
classifier basis: classifier_2025
4 ran · 9 no recorded run
FunctionStatusWhere it lives
_disable_dynamo_if_unsupported Ran
ran; our contract draft was wrong, not the code
automl/cpr/pytorch_cpr/adamcpr.py
code served (permissive licence) · get_code("77e74503b4a6ddf3")
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) 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
import functools

torch.manual_seed(0)

# Create a simple function to use as single_tensor_fn
def simple_tensor_fn(tensor):
    return tensor * 2

# Create a simple decorated function that will be wrapped
def sample_optimizer_step(params, grads, state_steps, capturable=False):
    """Sample function with state_steps parameter"""
    return sum(p + g for p, g in zip(params, grads))

# Create state_steps as a list of CPU tensors (not CUDA)
state_steps = [torch.tensor([1], dtype=torch.long)]

ARGS = {
    'single_tensor_fn': simple_tensor_fn
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
_get_capturable_supported_devices Ran
ran; our contract draft was wrong, not the code
automl/cpr/pytorch_cpr/adamcpr.py
code served (permissive licence) · get_code("8d0c76131beabbb9")
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) and returned list, ndim 1, dtype family float; no output shape is stated. 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 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 torch

torch.manual_seed(0)

ARGS = {
    'supports_xla': True,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — 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
group_parameters_for_cpr_optimizer Ran
ran; our contract draft was wrong, not the code
automl/cpr/pytorch_cpr/adamcpr.py
code served (permissive licence) · get_code("2f3d97790ecb08b8")
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 torch
import torch.nn as nn

torch.manual_seed(0)

# Create a small model with various parameter types
class TestModel(nn.Module):
    def __init__(self):
        super().__init__()
        # Linear layer with weight decay
        self.linear1 = nn.Linear(4, 3)
        # LayerNorm (should go to no_decay)
        self.norm = nn.LayerNorm(3)
        # Embedding (should go to no_decay when regularize_embedding=False)
        self.embed = nn.Embedding(10, 4)
        # BatchNorm (should go to no_decay)
        self.bn = nn.BatchNorm1d(3)
        # Another linear layer
        self.linear2 = nn.Linear(3, 2)
    
    def forward(self, x):
        x = self.linear1(x)
        x = self.norm(x)
        x = self.bn(x)
        x = self.linear2(x)
        return x

model = TestModel()

# Ensure all parameters require gradients
for param in model.parameters():
    param.requires_grad = True

ARGS = {
    'model': model,
    'regularize_embedding': 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 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
wd_group_named_parameters Ran
ran; our contract draft was wrong, not the code
automl/cpr/examples/train_cifar100_task.py
code served (permissive licence) · get_code("2d82e29243858703")
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) 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 torch
import torch.nn as nn

torch.manual_seed(0)

# Build a small model with various parameter types
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        # Linear layer - has weight and bias
        self.linear = nn.Linear(4, 3)
        # Conv2d layer - has weight and bias
        self.conv = nn.Conv2d(3, 2, kernel_size=3)
        # BatchNorm - should not have decay applied
        self.bn = nn.BatchNorm1d(2)
        # Embedding - weight should have decay
        self.embed = nn.Embedding(5, 4)
        # LayerNorm - should not have decay applied
        self.ln = nn.LayerNorm(4)
    
    def forward(self, x):
        return x

model = SimpleModel()

# Ensure all parameters require gradients
for param in model.parameters():
    param.requires_grad = True

weight_decay = 0.01

ARGS = {
    'model': model,
    'weight_decay': weight_decay,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — 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
AdamCPR Not yet run
unverified (harvested, no recorded run)
automl/cpr/pytorch_cpr/adamcpr.py
code served (permissive licence) · get_code("20c73d219b728426")
How it ran

no invocation recorded.

_multi_tensor_adamcpr Not yet run
unverified (harvested, no recorded run)
automl/cpr/pytorch_cpr/adamcpr.py
code served (permissive licence) · get_code("ee694b63b0bc4895")
How it ran

no invocation recorded.

_single_tensor_adamcpr Not yet run
unverified (harvested, no recorded run)
automl/cpr/pytorch_cpr/adamcpr.py
code served (permissive licence) · get_code("3e3e242295adb363")
How it ran

no invocation recorded.

adamcpr Not yet run
unverified (harvested, no recorded run)
automl/cpr/pytorch_cpr/adamcpr.py
code served (permissive licence) · get_code("a14f930855d7e136")
How it ran

no invocation recorded.

huber_update Not yet run
unverified (harvested, no recorded run)
automl/cpr/pytorch_cpr/adamcpr.py
code served (permissive licence) · get_code("5d028ca59f740e80")
How it ran

no invocation recorded.

l1_update Not yet run
unverified (harvested, no recorded run)
automl/cpr/pytorch_cpr/adamcpr.py
code served (permissive licence) · get_code("f5fe436ea316e71f")
How it ran

no invocation recorded.

l2_update Not yet run
unverified (harvested, no recorded run)
automl/cpr/pytorch_cpr/adamcpr.py
code served (permissive licence) · get_code("3babf488a0693b0b")
How it ran

no invocation recorded.

single_initilize_kappa Not yet run
unverified (harvested, no recorded run)
automl/cpr/pytorch_cpr/adamcpr.py
code served (permissive licence) · get_code("08254562e66103c1")
How it ran

no invocation recorded.

std_update Not yet run
unverified (harvested, no recorded run)
automl/cpr/pytorch_cpr/adamcpr.py
code served (permissive licence) · get_code("ddf970467b201f00")
How it ran

no invocation recorded.

Also present in other papers' code

Of 13 samples lifted from this paper's linked repositories, 13 carry a function identity and 0 do not. Of the 13 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 13 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

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

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

Repositories linked to this paper

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

Abstract

Regularization is a critical component in deep learning. The most commonly used approach, weight decay, applies a constant penalty coefficient uniformly across all parameters. This may be overly restrictive for some parameters, while insufficient for others. To address this, we present Constrained Parameter Regularization (CPR) as an alternative to traditional weight decay. Unlike the uniform application of a single penalty, CPR enforces an upper bound on a statistical measure, such as the L 2 -norm, of individual parameter matrices. Consequently, learning becomes a constraint optimization problem, which we tackle using an adaptation of the augmented Lagrangian method. CPR introduces only a minor runtime overhead and only requires setting an upper bound. We propose simple yet efficient mechanisms for initializing this bound, making CPR rely on no hyperparameter or one, akin to weight decay. Our empirical studies on computer vision and language modeling tasks demonstrate CPR's effectiveness. The results show that CPR can outperform traditional weight decay and increase performance in pre-training and fine-tuning.

For agents

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

get_harvested_code_for_paper("2311.09058")
get_code_for_paper("2311.09058")
have("2311.09058")

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

curl https://syntology.ai/api/ran/2311.09058.json

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

[![Syntology run record](https://syntology.ai/api/ran/2311.09058.svg)](https://syntology.ai/paper/2311.09058)

Connect an agent — have() is free.