SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2010.08034 · 2020

Overfitting or Underfitting? Understand Robustness Drop in Adversarial Training

arXiv · PDF · Open in the Atlas

Code that ran

4 of 5 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 3 have no contract verdict; 1 failure of our instrument; 1 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/2010.08034.json · badge: /api/ran/2010.08034.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
zichaoli/APART canonical (classifier label)
classifier basis: census
4 ran · 1 no recorded run
FunctionStatusWhere it lives
atta_aug Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
zichaoli/APART/adaptive_data_aug.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from HandingWangXDGroup/FGSM-PCO, is pointer only (licence: NONE) · get_code("eda67bbf007849b8")
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 HandingWangXDGroup/FGSM-PCO, 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.

atta_aug_trans Ran
ran on a synthesized input
zichaoli/APART/adaptive_data_aug.py
code served (permissive licence) · get_code("cd121dba2366ef98")
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 4, dtype family float, shape (2, 3, 32, 32) 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)

# Construct input_tensor: batch_size=2, channels=3, height=64, width=64
# (needs to be large enough to crop 32x32 from it)
input_tensor = torch.randn(2, 3, 64, 64)

# Construct transform_info dict with crop and flipped info
transform_info = {
    'crop': {
        'x': torch.tensor([0, 16]),  # x offsets for each item in batch
        'y': torch.tensor([0, 8]),   # y offsets for each item in batch
    },
    'flipped': torch.tensor([0, 1])  # flip flags for each item in batch
}

# Construct rst: output tensor, batch_size=2, channels=3, height=32, width=32
rst = torch.zeros(2, 3, 32, 32)

ARGS = {
    'input_tensor': input_tensor,
    'transform_info': transform_info,
    'rst': rst,
}
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 4, dtype family float, shape (2, 3, 32, 32) — shape (2, 3, 32, 32) 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
inverse_atta_aug Ran
ran on a synthesized input
zichaoli/APART/adaptive_data_aug.py
code served (permissive licence) · get_code("5a018a3a06e619c0")
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 4, dtype family float, shape (2, 3, 64, 64) 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)

# Construct source_tensor: batch_size=2, channels=3, height=64, width=64
# (needs to be large enough to hold 32x32 patches at various positions)
source_tensor = torch.randn(2, 3, 64, 64)

# Construct adv_tensor: batch_size=2, channels=3, height=32, width=32
# (the patch size used in the function)
adv_tensor = torch.randn(2, 3, 32, 32)

# Construct transform_info with crop and flipped information
transform_info = {
    'crop': {
        'x': torch.tensor([10, 20]),  # x offsets for each batch item
        'y': torch.tensor([15, 25]),  # y offsets for each batch item
    },
    'flipped': torch.tensor([0, 1]),  # flipped flag for each batch item (0 or 1)
}

ARGS = {
    'source_tensor': source_tensor,
    'adv_tensor': adv_tensor,
    'transform_info': transform_info,
}
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 4, dtype family float, shape (2, 3, 64, 64) — shape (2, 3, 64, 64) 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
preact20 Ran
ran on a synthesized input
zichaoli/APART/models/preresnet.py
code served (permissive licence) · get_code("bdaf671659de92ff")
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 PreActResNet. 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 = {
    'num_classes': 10
}
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 PreActResNet — a PreActResNet output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a PreActResNet 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
load_pading_training_data Not yet run
unverified (harvested, no recorded run)
zichaoli/APART/cifar_dataloader.py
code served (permissive licence) · get_code("0fcb1e999f8f49a0")
How it ran

no invocation recorded.

Also present in other papers' code

Of 5 samples lifted from this paper's linked repositories, 5 carry a function identity and 0 do not. Of the 5 distinct functions with an identity, 1 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 1 later) and 4 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
atta_aug 1 other paper
1 later
2407.12443 (2024-07)
Preventing Catastrophic Overfitting in Fast Adversarial Training: A Bi-level Optimization Perspective

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

Our goal is to understand why the robustness drops after conducting adversarial training for too long. Although this phenomenon is commonly explained as overfitting, our analysis suggest that its primary cause is perturbation underfitting. We observe that after training for too long, FGSM-generated perturbations deteriorate into random noise. Intuitively, since no parameter updates are made to strengthen the perturbation generator, once this process collapses, it could be trapped in such local optima. Also, sophisticating this process could mostly avoid the robustness drop, which supports that this phenomenon is caused by underfitting instead of overfitting. In the light of our analyses, we propose APART, an adaptive adversarial training framework, which parameterizes perturbation generation and progressively strengthens them. Shielding perturbations from underfitting unleashes the potential of our framework. In our experiments, APART provides comparable or even better robustness than PGD-10, with only about 1/4 of its computational cost.

For agents

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

get_harvested_code_for_paper("2010.08034")
get_code_for_paper("2010.08034")
have("2010.08034")

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

curl https://syntology.ai/api/ran/2010.08034.json

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

[![Syntology run record](https://syntology.ai/api/ran/2010.08034.svg)](https://syntology.ai/paper/2010.08034)

Connect an agent — have() is free.