SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2007.06346 · ICML · 2021

Whitening for Self-Supervised Representation Learning

Aliaksandr Siarohin, Nicu Sebe, Aleksandr Ermolov, Enver Sangineto

arXiv · PDF · Open in the Atlas

Code that ran

11 of 13 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (3 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 0 violated it and 4 have no contract verdict; 6 failures of our instrument; 2 no recorded run), measured 2026-11-21; 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/2007.06346.json · badge: /api/ran/2007.06346.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
htdt/self-supervised canonical (classifier label)
classifier basis: classifier_2025
1 ran · 0 no recorded run
dr4thmos/solo-learn-radio community (classifier label)
classifier basis not recorded on the link
8 ran · 1 no recorded run
trungpx/solo-learn-1214 community
Papers with Code archive
1 ran · 0 no recorded run
vturrisi/solo-learn community
Papers with Code archive
1 ran · 0 no recorded run
winci-ai/cw-rgp community
Papers with Code archive
0 ran · 1 no recorded run
FunctionStatusWhere it lives
Whitening2d Ran
ran on a synthesized input · constructed an object; no result was computed
htdt/self-supervised/methods/whitening.py
pointer only (licence: NONE) · get_code("2fd07073535ad61b")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, htdt/self-supervised 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.

Whitening2d Ran
ran on a synthesized input · constructed an object; no result was computed
trungpx/solo-learn-1214/solo/utils/whitening.py
code served (permissive licence) · get_code("8e1c3896cc666ecd")
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 an instance of its own class, Whitening2d: the run constructed the object and did not call it, so no result was computed. 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 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
from torch.cuda.amp import custom_fwd
from torch.nn.functional import conv2d

torch.manual_seed(0)

class Whitening2d(nn.Module):

    def __init__(self, output_dim: int, eps: float=0.0):
        """Layer that computes hard whitening for W-MSE using the Cholesky decomposition.

        Args:
            output_dim (int): number of dimension of projected features.
            eps (float, optional): eps for numerical stability in Cholesky decomposition. Defaults
                to 0.0.
        """
        super(Whitening2d, self).__init__()
        self.output_dim = output_dim
        self.eps = eps

    @custom_fwd(cast_inputs=torch.float32)
    def forward(self, x: torch.Tensor) -> torch.Tensor:
        """Performs whitening using the Cholesky decomposition.

        Args:
            x (torch.Tensor): a batch or slice of projected features.

        Returns:
            torch.Tensor: a batch or slice of whitened features.
        """
        x = x.unsqueeze(2).unsqueeze(3)
        m = x.mean(0).view(self.output_dim, -1).mean(-1).view(1, -1, 1, 1)
        xn = x - m
        T = xn.permute(1, 0, 2, 3).contiguous().view(self.output_dim, -1)
        f_cov = torch.mm(T, T.permute(1, 0)) / (T.shape[-1] - 1)
        eye = torch.eye(self.output_dim).type(f_cov.type())
        f_cov_shrinked = (1 - self.eps) * f_cov + self.eps * eye
        inv_sqrt = torch.triangular_solve(eye, torch.cholesky(f_cov_shrinked), upper=False)[0]
        inv_sqrt = inv_sqrt.contiguous().view(self.output_dim, self.output_dim, 1, 1)
        decorrelated = conv2d(xn, inv_sqrt)
        return decorrelated.squeeze(2).squeeze(2)

output_dim = 4
eps = 0.01
x = torch.randn(2, 4)

whitening_layer = Whitening2d(output_dim=output_dim, eps=eps)

ARGS = {
    "output_dim": output_dim,
    "eps": eps,
}
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 Whitening2d — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a Whitening2d 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
Whitening2d Ran
ran on a synthesized input · constructed an object; no result was computed
vturrisi/solo-learn/solo/utils/whitening.py
code served (permissive licence) · get_code("5ec9dc11fb54ff22")
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 an instance of its own class, Whitening2d: the run constructed the object and did not call it, so no result was computed. 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 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)

# Whitening2d expects:
# - output_dim: int - the number of output dimensions
# - eps: float - epsilon for numerical stability (optional, defaults to 0.0)

# Looking at the forward method:
# - x is expected to be a tensor of shape (batch_size, output_dim, ...)
# - The method unsqueezes it to add spatial dimensions if needed
# - It computes mean, covariance, and Cholesky decomposition
# - Final output is squeezed back

# Create a small instance with output_dim=4 and eps=0.01

ARGS = {
    'output_dim': 4,
    'eps': 0.01,
}
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 Whitening2d — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a Whitening2d 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
barlow_loss_func Ran
ran; our contract draft was wrong, not the code
dr4thmos/solo-learn-radio/solo/losses/barlow.py
code served (permissive licence) · get_code("132a949a804eb81e")
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 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
import torch.nn as nn

torch.manual_seed(0)

# Create small tensors for z1 and z2
# N=2 (batch size), D=4 (feature dimension)
z1 = torch.randn(2, 4)
z2 = torch.randn(2, 4)

# Set standard hyperparameters
lamb = 0.005
scale_loss = 0.025

ARGS = {
    "z1": z1,
    "z2": z2,
    "lamb": lamb,
    "scale_loss": scale_loss,
}
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 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
byol_loss_func Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
dr4thmos/solo-learn-radio/solo/losses/byol.py
pointer only (licence: MIT); the stored copy get_code serves, from PKU-ML/nlb, is pointer only (licence: NONE) · get_code("273532e06cb3967e")
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 PKU-ML/nlb, 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.

mae_loss_func Ran
ran; our contract draft was wrong, not the code
dr4thmos/solo-learn-radio/solo/losses/mae.py
code served (permissive licence) · get_code("a07d9a7525702b80")
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 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

torch.manual_seed(0)

# Create valid arguments for mae_loss_func
# imgs: [N, 3, H, W] - batch of 2, 3 channels, 4x4 spatial dims
imgs = torch.randn(2, 3, 4, 4)

# patch_size: int - must divide evenly into H and W
patch_size = 2

# pred: [N, Tokens, pixels * pixels * 3]
# With patch_size=2 and H=W=4, we have h=w=2, so Tokens=4
# pixels*pixels*3 = 2*2*3 = 12
pred = torch.randn(2, 4, 12)

# mask: [N, Tokens] - binary mask where 1 means masked
# With 4 tokens, mask shape is [2, 4]
mask = torch.tensor([[1, 0, 1, 0], [1, 1, 0, 0]], dtype=torch.float32)

# norm_pix_loss: bool
norm_pix_loss = True

ARGS = {
    "imgs": imgs,
    "pred": pred,
    "mask": mask,
    "patch_size": patch_size,
    "norm_pix_loss": norm_pix_loss,
}
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 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
mocov2plus_loss_func Ran
ran; our contract draft was wrong, not the code
dr4thmos/solo-learn-radio/solo/losses/mocov2plus.py
code served (permissive licence) · get_code("4dd7c3cff0150cb9")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) and returned Tensor, 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
import torch.nn.functional as F

torch.manual_seed(0)

# Create query: NxD tensor (batch_size=2, embedding_dim=4)
query = torch.randn(2, 4)
query = F.normalize(query, dim=1)  # Normalize as typical in contrastive learning

# Create key: NxD tensor (batch_size=2, embedding_dim=4)
key = torch.randn(2, 4)
key = F.normalize(key, dim=1)

# Create queue: DxK tensor (embedding_dim=4, queue_size=6)
# Note: queue is transposed (D x K) for efficient einsum computation
queue = torch.randn(4, 6)
queue = F.normalize(queue, dim=0)  # Normalize along embedding dimension

# Temperature parameter
temperature = 0.1

ARGS = {
    'query': query,
    'key': key,
    'queue': queue,
    'temperature': temperature
}
Setup block drafted by
a language model, Claude Sonnet 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 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
patchify Ran
ran; our contract draft was wrong, not the code
dr4thmos/solo-learn-radio/solo/losses/mae.py
code served (permissive licence) · get_code("3e6a557bf1ad83d9")
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 3, dtype family float, shape (2, 4, 12) 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)

# Create a small batch of images
# Shape: [N=2, 3, H=4, W=4] where H and W are divisible by patch_size
imgs = torch.randn(2, 3, 4, 4)

# Patch size that divides evenly into image dimensions
patch_size = 2

ARGS = {
    "imgs": imgs,
    "patch_size": patch_size,
}
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 3, dtype family float, shape (2, 4, 12) — shape (2, 4, 12) 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
ressl_loss_func Ran
ran, honoured the contract we drafted
dr4thmos/solo-learn-radio/solo/losses/ressl.py
code served (permissive licence) · get_code("779f0ded23fd8106")
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 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
import torch.nn.functional as F

torch.manual_seed(0)

# Create tiny tensors for the function
batch_size = 2
embedding_dim = 4
queue_size = 3

# q: NxD tensor (queries from view 1)
q = torch.randn(batch_size, embedding_dim)

# k: NxD tensor (keys from view 2)
k = torch.randn(batch_size, embedding_dim)

# queue: KxD tensor (past elements for contrastive loss)
queue = torch.randn(queue_size, embedding_dim)

# Temperature parameters
temperature_q = 0.1
temperature_k = 0.04

ARGS = {
    "q": q,
    "k": k,
    "queue": queue,
    "temperature_q": temperature_q,
    "temperature_k": temperature_k,
}
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 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
simsiam_loss_func Ran
ran on a synthesized input
dr4thmos/solo-learn-radio/solo/losses/simsiam.py
code served (permissive licence) · get_code("cd64814bdf3905fa")
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 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
import torch.nn.functional as F

torch.manual_seed(0)

# Create small tensors with distinct dimensions
# p and z should have the same shape: (batch_size, feature_dim)
batch_size = 2
feature_dim = 4

p = torch.randn(batch_size, feature_dim)
z = torch.randn(batch_size, feature_dim)

simplified = True

ARGS = {
    "p": p,
    "z": z,
    "simplified": simplified,
}
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 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
swav_loss_func Ran
ran; our contract draft was wrong, not the code
dr4thmos/solo-learn-radio/solo/losses/swav.py
code served (permissive licence) · get_code("6b191c39aa0c6668")
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 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
import numpy as np

torch.manual_seed(0)

# Create sample predictions and assignments for SwAV loss
# The function expects lists of tensors where each tensor is NxC (batch_size x num_clusters)
batch_size = 2
num_clusters = 4
num_views = 3

# Create list of predictions (one per view)
preds = [
    torch.randn(batch_size, num_clusters),
    torch.randn(batch_size, num_clusters),
    torch.randn(batch_size, num_clusters),
]

# Create list of assignments (one per view)
# Assignments should be probability-like (sum to 1 along dim 1)
assignments = []
for _ in range(num_views):
    a = torch.rand(batch_size, num_clusters)
    a = a / a.sum(dim=1, keepdim=True)  # Normalize to sum to 1
    assignments.append(a)

temperature = 0.1

ARGS = {
    "preds": preds,
    "assignments": assignments,
    "temperature": temperature,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, 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
Whitening2d Not yet run
unverified (harvested, no recorded run)
winci-ai/cw-rgp/ImageNet/methods/whitening.py
pointer only (licence: NONE) · get_code("d09f66fe441f6036")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, winci-ai/cw-rgp 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. No invocation recorded.

deepclusterv2_loss_func Not yet run
unverified (harvested, no recorded run)
dr4thmos/solo-learn-radio/solo/losses/deepclusterv2.py
pointer only (licence: MIT); the stored copy get_code serves, from PKU-ML/nlb, is pointer only (licence: NONE) · get_code("ce432f2e9653c7a3")
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 PKU-ML/nlb, 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. 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, 8 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 8 later) and 5 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
barlow_loss_func 12 other papers
12 later
2108.01775 (2021-08)
Solo-learn: A Library of Self-supervised Methods for Visual Representation Learning
2112.15022 (2021-12)
Continually Learning Self-Supervised Representations with Projected Functional Regularization
and 9 more not named here
deepclusterv2_loss_func 10 other papers
10 later
2108.01775 (2021-08)
Solo-learn: A Library of Self-supervised Methods for Visual Representation Learning
2112.15022 (2021-12)
Continually Learning Self-Supervised Representations with Projected Functional Regularization
and 7 more not named here
byol_loss_func 6 other papers
6 later
2204.00616 (2022-04)
Simplicial Embeddings in Self-Supervised Learning and Downstream Classification
2310.02903 (2023-10)
FroSSL: Frobenius Norm Minimization for Efficient Multiview Self-Supervised Learning
2403.12448 (2024-03)
Do Generated Data Always Help Contrastive Learning?
and 3 more not named here
mocov2plus_loss_func 5 other papers
1 dated earlier, 4 later
2003.04297 (2020-03)
Improved Baselines with Momentum Contrastive Learning
2403.12448 (2024-03)
Do Generated Data Always Help Contrastive Learning?
2403.12459 (2024-03)
Non-negative Contrastive Learning
and 2 more not named here
ressl_loss_func 5 other papers
5 later
2107.09282 (2021-07)
ReSSL: Relational Self-Supervised Learning with Weak Augmentation
2108.01775 (2021-08)
Solo-learn: A Library of Self-supervised Methods for Visual Representation Learning
2112.15022 (2021-12)
Continually Learning Self-Supervised Representations with Projected Functional Regularization
and 2 more not named here
mae_loss_func 4 other papers
4 later
2310.02903 (2023-10)
FroSSL: Frobenius Norm Minimization for Efficient Multiview Self-Supervised Learning
2403.12448 (2024-03)
Do Generated Data Always Help Contrastive Learning?
2403.12459 (2024-03)
Non-negative Contrastive Learning
and 1 more not named here
patchify 4 other papers
4 later
2310.02903 (2023-10)
FroSSL: Frobenius Norm Minimization for Efficient Multiview Self-Supervised Learning
2403.12448 (2024-03)
Do Generated Data Always Help Contrastive Learning?
2403.12459 (2024-03)
Non-negative Contrastive Learning
and 1 more not named here
swav_loss_func 1 other paper
1 later
2410.04959 (2024-10)
Collapse-Proof NON-CONTRASTIVE SELF-SUPERVISED LEARNING

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 2026-11-21.

Repositories linked to this paper

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

Abstract

Most of the current self-supervised representation learning (SSL) methods are based on the contrastive loss and the instance-discrimination task, where augmented versions of the same image instance ("positives") are contrasted with instances extracted from other images ("negatives"). For the learning to be effective, many negatives should be compared with a positive pair, which is computationally demanding. In this paper, we propose a different direction and a new loss function for SSL, which is based on the whitening of the latentspace features. The whitening operation has a "scattering" effect on the batch samples, avoiding degenerate solutions where all the sample representations collapse to a single point. Our solution does not require asymmetric networks and it is conceptually simple. Moreover, since negatives are not needed, we can extract multiple positive pairs from the same image instance. The source code of the method and of all the experiments is available at: https://github.com/htdt/ self-supervised.

For agents

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

get_harvested_code_for_paper("2007.06346")
get_code_for_paper("2007.06346")
have("2007.06346")

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

curl https://syntology.ai/api/ran/2007.06346.json

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

[![Syntology run record](https://syntology.ai/api/ran/2007.06346.svg)](https://syntology.ai/paper/2007.06346)

Connect an agent — have() is free.