SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2411.00899 · NeurIPS · 2024

Certified Robustness for Deep Equilibrium Models via Serialized Random Smoothing

Xiaorui Liu, Han Xu, Weizhi Gao

arXiv · PDF · Open in the Atlas

Code that ran

12 of 14 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 0 have no contract verdict; 9 failures of our instrument; 2 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/2411.00899.json · badge: /api/ran/2411.00899.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
WeizhiGao/Serialized-Randomized-Smoothing canonical (classifier label)
classifier basis: census
12 ran · 2 no recorded run
FunctionStatusWhere it lives
conv3x3 Ran
ran; our contract draft was wrong, not the code
WeizhiGao/Serialized-Randomized-Smoothing/SRS/code/archs/cifar_resnet.py
pointer only (licence: MIT); the stored copy get_code serves, from lixiaotong97/DSU, is pointer only (licence: NONE) · get_code("fac5364e2f53c6db")
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 lixiaotong97/DSU, 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.

conv3x3 Ran
ran; our contract draft was wrong, not the code
WeizhiGao/Serialized-Randomized-Smoothing/DEQ/lib/layer_utils.py
code served (permissive licence) · get_code("bc359191fdd6e179")
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 Conv2d. 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_planes": 3,
    "out_planes": 64,
    "stride": 1,
    "bias": 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 Conv2d — a Conv2d output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Conv2d 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
embedded_dropout Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
WeizhiGao/Serialized-Randomized-Smoothing/DEQ/lib/optimizations.py
pointer only (licence: MIT); the stored copy get_code serves, from landrarwolf/CDEQ, is pointer only (licence: NONE) · get_code("df1c4e8a21c5f2ef")
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 landrarwolf/CDEQ, 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.

get_num_classes Ran
ran, honoured the contract we drafted
WeizhiGao/Serialized-Randomized-Smoothing/SRS/code/datasets.py
code served (permissive licence) · get_code("8c99213918181c27")
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

torch.manual_seed(0)

ARGS = {
    'dataset': 'cifar10'
}
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
line_search Ran
ran; our contract draft was wrong, not the code
WeizhiGao/Serialized-Randomized-Smoothing/DEQ/lib/solvers.py
code served (permissive licence) · get_code("cf5710371bcfb0ab")
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 tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# Create a small tensor for x0
x0 = torch.randn(4, requires_grad=True)

# Create update direction
update = torch.randn(4, requires_grad=True)

# Create initial gradient g0
g0 = torch.randn(4)

# Define a simple gradient function g that takes x_est and returns a gradient
def g(x_est):
    """
    Simple gradient function that returns a gradient vector.
    Takes an estimated x and returns a gradient of same shape.
    """
    return torch.randn(4)

ARGS = {
    "update": update,
    "x0": x0,
    "g0": g0,
    "g": g,
    "nstep": 0,
    "on": True,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
list2vec Ran
ran, honoured the contract we drafted
WeizhiGao/Serialized-Randomized-Smoothing/DEQ/lib/layer_utils.py
pointer only (licence: MIT); the stored copy get_code serves, from landrarwolf/CDEQ, is pointer only (licence: NONE) · get_code("e193f0d97cdfd598")
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 landrarwolf/CDEQ, 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.

matrix_diag Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
WeizhiGao/Serialized-Randomized-Smoothing/DEQ/lib/optimizations.py
pointer only (licence: MIT); the stored copy get_code serves, from landrarwolf/CDEQ, is pointer only (licence: NONE) · get_code("f06524d1a24b825a")
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 landrarwolf/CDEQ, 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.

power_method Ran
ran; our contract draft was wrong, not the code
WeizhiGao/Serialized-Randomized-Smoothing/DEQ/lib/jacobian.py
code served (permissive licence) · get_code("3c120cc50bc8c55f")
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 tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

# Create a simple function that takes z0 as input and produces f0 as output
# We need z0 to require gradients so that torch.autograd.grad can compute gradients
z0 = torch.randn(2, 4, requires_grad=True)

# Create a simple neural network to define the function f
class SimpleNet(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 4)
    
    def forward(self, x):
        return self.linear(x)

net = SimpleNet()

# Compute f0 by passing z0 through the network
f0 = net(z0)

ARGS = {
    "f0": f0,
    "z0": z0,
    "n_iters": 200
}
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 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
rmatvec Ran
ran; our contract draft was wrong, not the code
WeizhiGao/Serialized-Randomized-Smoothing/DEQ/lib/solvers.py
code served (permissive licence) · get_code("968a8f446ad3c5b1")
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 Tensor, ndim 3, dtype family float, shape (2, 3, 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)

# Analyze the function:
# - part_Us: used in einsum 'bij, bijd -> bd', so shape (b, i, j, d)
# - part_VTs: used in einsum 'bd, bdij -> bij', so shape (b, d, i, j)
# - x: used in einsum 'bij, bijd -> bd', so shape (b, i, j)

# Create small tensors with distinct dimensions
batch_size = 2
i_dim = 3
j_dim = 4
d_dim = 5

part_Us = torch.randn(batch_size, i_dim, j_dim, d_dim)
part_VTs = torch.randn(batch_size, d_dim, i_dim, j_dim)
x = torch.randn(batch_size, i_dim, j_dim)

ARGS = {
    "part_Us": part_Us,
    "part_VTs": part_VTs,
    "x": x,
}
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 Tensor, ndim 3, dtype family float, shape (2, 3, 4) — shape (2, 3, 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
scalar_search_armijo Ran
ran, honoured the contract we drafted
WeizhiGao/Serialized-Randomized-Smoothing/DEQ/lib/solvers.py
code served (permissive licence) · get_code("b2b5022f654ee4d3")
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 tuple, 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

torch.manual_seed(0)

# The function expects:
# - phi: a callable that takes a scalar alpha and returns a scalar value
# - phi0: a scalar (initial function value)
# - derphi0: a scalar (initial derivative value)
# - c1: a scalar (default 0.0001)
# - alpha0: a scalar (default 1)
# - amin: a scalar (default 0)

# Create a simple quadratic function as phi
# phi(alpha) = (alpha - 0.5)^2 + 0.1
def phi(alpha):
    return (alpha - 0.5) ** 2 + 0.1

# Evaluate at alpha=1 to get phi0
phi0 = phi(1.0)

# Derivative of phi at alpha=1: d/d_alpha[(alpha - 0.5)^2 + 0.1] = 2(alpha - 0.5)
# At alpha=1: 2(1 - 0.5) = 1.0
derphi0 = 1.0

ARGS = {
    "phi": phi,
    "phi0": phi0,
    "derphi0": derphi0,
    "c1": 0.0001,
    "alpha0": 1.0,
    "amin": 0.0,
}
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 tuple, ndim 1, dtype family float — a tuple 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
vec2list Ran
ran; our contract draft was wrong, not the code
WeizhiGao/Serialized-Randomized-Smoothing/DEQ/lib/layer_utils.py
pointer only (licence: MIT); the stored copy get_code serves, from landrarwolf/CDEQ, is pointer only (licence: NONE) · get_code("bfc68a2da5929c5b")
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 landrarwolf/CDEQ, 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.

weight_norm Ran
ran; our contract draft was wrong, not the code
WeizhiGao/Serialized-Randomized-Smoothing/DEQ/lib/optimizations.py
pointer only (licence: MIT); the stored copy get_code serves, from landrarwolf/CDEQ, is pointer only (licence: NONE) · get_code("4bf00e788db9f349")
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 landrarwolf/CDEQ, 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.

get_dataset Not yet run
unverified (harvested, no recorded run)
WeizhiGao/Serialized-Randomized-Smoothing/SRS/code/datasets.py
code served (permissive licence) · get_code("667ac97f3d05bca6")
How it ran

no invocation recorded.

jac_loss_estimate Not yet run
unverified (harvested, no recorded run)
WeizhiGao/Serialized-Randomized-Smoothing/DEQ/lib/jacobian.py
code served (permissive licence) · get_code("3d07af47a6ba65bd")
How it ran

no invocation recorded.

Also present in other papers' code

Of 14 samples lifted from this paper's linked repositories, 14 carry a function identity and 0 do not. Of the 14 distinct functions with an identity, 14 are also present in other papers whose code we hold (11 with at least one paper dated earlier, 1 the same month, 12 later) and 0 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
conv3x3 686 other papers
636 dated earlier, 5 the same month, 28 later, 17 same year, order not determinable · 31 only as an assembly brick
1406.2661 (2014-06)
Generative Adversarial Networks
1409.0575 (2014-09)
ImageNet Large Scale Visual Recognition Challenge
1412.6572 (2014-12)
Explaining and Harnessing Adversarial Examples
and 683 more not named here
scalar_search_armijo 13 other papers
11 dated earlier, 2 later · 4 only as an assembly brick
2006.08656 (2020-06)
Multiscale Deep Equilibrium Models
2106.00553 (2021-06)
SHINE: SHaring the INverse Estimate from the forward pass for bi-level optimization and implicit models
2106.08748 (2021-06)
Input Invex Neural Network
and 10 more not named here
line_search 11 other papers
9 dated earlier, 2 later · 3 only as an assembly brick
2006.08656 (2020-06)
Multiscale Deep Equilibrium Models
2106.08748 (2021-06)
Input Invex Neural Network
2109.14247 (2021-09)
Training Feedback Spiking Neural Networks by Implicit Differentiation on the Equilibrium State
and 8 more not named here
rmatvec 10 other papers
8 dated earlier, 2 later · 5 only as an assembly brick
2006.08656 (2020-06)
Multiscale Deep Equilibrium Models
2106.00553 (2021-06)
SHINE: SHaring the INverse Estimate from the forward pass for bi-level optimization and implicit models
2109.14247 (2021-09)
Training Feedback Spiking Neural Networks by Implicit Differentiation on the Equilibrium State
and 7 more not named here
conv3x3 8 other papers
7 dated earlier, 1 later
1512.03385 (2015-12)
Deep Residual Learning for Image Recognition
1805.11046 (2018-05)
Scalable Methods for 8-bit Training of Neural Networks
2001.07645 (2020-01)
SAUNet: Shape Attentive U-Net for Interpretable Medical Image Segmentation
and 5 more not named here
weight_norm 5 other papers
4 dated earlier, 1 later
1810.06682 (2018-10)
Trellis Networks for Sequence Modeling
2006.08656 (2020-06)
Multiscale Deep Equilibrium Models
2111.05177 (2021-11)
On Training Implicit Models
and 2 more not named here
get_num_classes 4 other papers
4 dated earlier
1906.04584 (2019-06)
Provably Robust Deep Learning via Adversarially Trained Smoothed Classifiers
2002.03517 (2020-02)
Random Smoothing Might be Unable to Certify $\ell_\infty$ Robustness for High-Dimensional Images
2206.10550 (2022-06)
(Certified!!) Adversarial Robustness for Free!
and 1 more not named here
jac_loss_estimate 3 other papers
2 dated earlier, 1 later
2106.14342 (2021-06)
Stabilizing Equilibrium Models by Jacobian Regularization
2304.11663 (2023-04)
Efficient Training of Deep Equilibrium Models
2602.03024 (2026-02)
Consistency Deep Equilibrium Models
power_method 3 other papers
2 dated earlier, 1 later
2106.14342 (2021-06)
Stabilizing Equilibrium Models by Jacobian Regularization
2304.11663 (2023-04)
Efficient Training of Deep Equilibrium Models
2602.03024 (2026-02)
Consistency Deep Equilibrium Models
embedded_dropout 2 other papers
1 dated earlier, 1 later
1711.03953 (2017-11)
Breaking the Softmax Bottleneck: A High-Rank RNN Language Model
2602.03024 (2026-02)
Consistency Deep Equilibrium Models
get_dataset 2 other papers
2 dated earlier
2206.10550 (2022-06)
(Certified!!) Adversarial Robustness for Free!
2308.14333 (2023-08)
DiffSmooth: Certifiably Robust Learning via Diffusion Models and Local Smoothing
list2vec 1 other paper
1 later
2602.03024 (2026-02)
Consistency Deep Equilibrium Models
matrix_diag 1 other paper
1 later
2602.03024 (2026-02)
Consistency Deep Equilibrium Models
vec2list 1 other paper
1 later
2602.03024 (2026-02)
Consistency Deep Equilibrium Models

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

Implicit models such as Deep Equilibrium Models (DEQs) have emerged as promising alternative approaches for building deep neural networks. Their certified robustness has gained increasing research attention due to security concerns. Existing certified defenses for DEQs employing deterministic certification methods such as interval bound propagation and Lipschitz-bounds can not certify on large-scale datasets. Besides, they are also restricted to specific forms of DEQs. In this paper, we provide the first randomized smoothing certified defense for DEQs to solve these limitations. Our study reveals that simply applying randomized smoothing to certify DEQs provides certified robustness generalized to large-scale datasets but incurs extremely expensive computation costs. To reduce computational redundancy, we propose a novel Serialized Randomized Smoothing (SRS) approach that leverages historical information. Additionally, we derive a new certified radius estimation for SRS to theoretically ensure the correctness of our algorithm. Extensive experiments and ablation studies on image recognition demonstrate that our algorithm can significantly accelerate the certification of DEQs by up to 7x almost without sacrificing the certified accuracy. Our code is available at https://github.com/WeizhiGao/Serialized-Randomized-Smoothing.

For agents

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

get_harvested_code_for_paper("2411.00899")
get_code_for_paper("2411.00899")
have("2411.00899")

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

curl https://syntology.ai/api/ran/2411.00899.json

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

[![Syntology run record](https://syntology.ai/api/ran/2411.00899.svg)](https://syntology.ai/paper/2411.00899)

Connect an agent — have() is free.