SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1901.09960 · 2019

Using Pre-Training Can Improve Model Robustness and Uncertainty

arXiv · PDF · Open in the Atlas

Code that ran

7 of 7 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, 1 violated it and 2 have no contract verdict; 4 failures of our instrument; 0 no recorded run); 1 of them was also run on the host, not in the jail, by the peer-agreement lane that set its level, 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/1901.09960.json · badge: /api/ran/1901.09960.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
hendrycks/pre-training canonical (classifier label)
classifier basis not recorded on the link
7 ran · 0 no recorded run
FunctionStatusWhere it lives
cosine_annealing Ran
ran, violated the contract we drafted
hendrycks/pre-training/downsampled_train/baseline.py
code served (permissive licence) · get_code("05eda95f9800a1e9")
How it ran

Ran in Syntology's jail on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned float64, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. The comparison behind this sample's level is a different run: the peer-agreement lane ran it again on the host, not in the jail (report dated 2026-09-04T22:12:48Z). 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 numpy as np

ARGS = {
    "step": 5,
    "total_steps": 100,
    "lr_max": 0.1,
    "lr_min": 0.001,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float64, 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
flip_labels_C Ran
ran on a synthesized input
hendrycks/pre-training/robustness/label_corruption/load_corrupted_ours_hierarchical.py
code served (permissive licence) · get_code("74fa122d15c0f3ce")
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 ndarray, ndim 2, dtype family float, shape (5, 5) 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 numpy as np

ARGS = {
    "corruption_prob": 0.3,
    "num_classes": 5,
    "seed": 42,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family float, shape (5, 5) — shape (5, 5) 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
make_layers Ran
ran; our contract draft was wrong, not the code
hendrycks/pre-training/downsampled_train/models/allconv.py
code served (permissive licence) · get_code("72410f1136cc7250")
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 Sequential. 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 make_layers function takes a cfg parameter which is a list/sequence
# of configuration values that can be:
# - 'Md' for MaxPool2d + Dropout
# - 'A' for AvgPool2d
# - 'NIN' for 1x1 conv layer
# - 'nopad' for 3x3 conv without padding
# - integers for output channels (3x3 conv with padding)

cfg = [64, 'NIN', 'Md', 128, 'nopad', 'A']

ARGS = {
    'cfg': cfg
}
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 Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 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
normalize_l2 Ran
ran; our contract draft was wrong, not the code
hendrycks/pre-training/robustness/adversarial/attacks.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from joelsimonoff/gram-adversary-detection, is pointer only (licence: NONE) · get_code("b16e920816efe253")
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 joelsimonoff/gram-adversary-detection, 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.

tensor_clamp Ran
ran; our contract draft was wrong, not the code
hendrycks/pre-training/robustness/adversarial/attacks.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from joelsimonoff/gram-adversary-detection, is pointer only (licence: NONE) · get_code("5b640b32a444ae38")
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 joelsimonoff/gram-adversary-detection, 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.

tensor_clamp_l2 Ran
ran; our contract draft was wrong, not the code
hendrycks/pre-training/robustness/adversarial/attacks.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from joelsimonoff/gram-adversary-detection, is pointer only (licence: NONE) · get_code("63b4423b18c974ad")
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 joelsimonoff/gram-adversary-detection, 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.

uniform_mix_C Ran
ran on a synthesized input
hendrycks/pre-training/robustness/label_corruption/load_corrupted_ours_hierarchical.py
code served (permissive licence) · get_code("ba6749b25e502f54")
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 ndarray, ndim 2, dtype family float, shape (5, 5) 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 numpy as np

# Construct valid arguments for uniform_mix_C
mixing_ratio = 0.3
num_classes = 5

ARGS = {
    "mixing_ratio": mixing_ratio,
    "num_classes": num_classes
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family float, shape (5, 5) — shape (5, 5) 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

Also present in other papers' code

Of 7 samples lifted from this paper's linked repositories, 7 carry a function identity and 0 do not. Of the 7 distinct functions with an identity, 7 are also present in other papers whose code we hold (5 with at least one paper dated earlier, 0 the same month, 7 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
cosine_annealing 9 other papers
9 later
2003.09711 (2020-03)
Robust Out-of-distribution Detection for Neural Networks
2108.11941 (2021-08)
Semantically Coherent Out-of-Distribution Detection
2110.01200 (2021-10)
AASIST: Audio Anti-Spoofing using Integrated Spectro-Temporal Graph Attention Networks
and 6 more not named here
uniform_mix_C 6 other papers
1 dated earlier, 5 later
1802.05300 (2018-02)
Using Trusted Data to Train Deep Networks on Labels Corrupted by Severe Noise
1902.07379 (2019-02)
Meta-Weight-Net: Learning an Explicit Mapping For Sample Weighting
1911.03809 (2019-11)
Meta Label Correction for Noisy Label Learning
and 3 more not named here
flip_labels_C 5 other papers
1 dated earlier, 4 later
1802.05300 (2018-02)
Using Trusted Data to Train Deep Networks on Labels Corrupted by Severe Noise
1902.07379 (2019-02)
Meta-Weight-Net: Learning an Explicit Mapping For Sample Weighting
1911.03809 (2019-11)
Meta Label Correction for Noisy Label Learning
and 2 more not named here
tensor_clamp 4 other papers
1 dated earlier, 3 later
1508.06576 (2015-08)
A Neural Algorithm of Artistic Style
1906.12340 (2019-06)
Using Self-Supervised Learning Can Improve Model Robustness and Uncertainty
1908.08016 (2019-08)
Testing Robustness Against Unforeseen Adversaries
and 1 more not named here
make_layers 3 other papers
3 later
1906.12340 (2019-06)
Using Self-Supervised Learning Can Improve Model Robustness and Uncertainty
2010.08001 (2020-10)
Maximum-Entropy Adversarial Data Augmentation for Improved Generalization and Robustness
2406.01494 (2024-06)
Robust Classification by Coupling Data Mollification with Label Smoothing
normalize_l2 3 other papers
1 dated earlier, 2 later
1508.06576 (2015-08)
A Neural Algorithm of Artistic Style
1906.12340 (2019-06)
Using Self-Supervised Learning Can Improve Model Robustness and Uncertainty
1912.12510 (2019-12)
Detecting Out-of-Distribution Examples with In-distribution Examples and Gram Matrices
tensor_clamp_l2 3 other papers
1 dated earlier, 2 later
1508.06576 (2015-08)
A Neural Algorithm of Artistic Style
1906.12340 (2019-06)
Using Self-Supervised Learning Can Improve Model Robustness and Uncertainty
1912.12510 (2019-12)
Detecting Out-of-Distribution Examples with In-distribution Examples and Gram Matrices

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

He et al. (2018) have called into question the utility of pre-training by showing that training from scratch can often yield similar performance to pre-training. We show that although pre-training may not improve performance on traditional classification metrics, it improves model robustness and uncertainty estimates. Through extensive experiments on adversarial examples, label corruption, class imbalance, out-of-distribution detection, and confidence calibration, we demonstrate large gains from pre-training and complementary effects with task-specific methods. We introduce adversarial pre-training and show approximately a 10% absolute improvement over the previous state-of-the-art in adversarial robustness. In some cases, using pre-training without task-specific methods also surpasses the state-of-the-art, highlighting the need for pre-training when evaluating future methods on robustness and uncertainty tasks.

For agents

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

get_harvested_code_for_paper("1901.09960")
get_code_for_paper("1901.09960")
have("1901.09960")

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

curl https://syntology.ai/api/ran/1901.09960.json

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

[![Syntology run record](https://syntology.ai/api/ran/1901.09960.svg)](https://syntology.ai/paper/1901.09960)

Connect an agent — have() is free.