SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2011.13377 · 2020

How Well Do Self-Supervised Models Transfer?

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; 1 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 3 failures of our instrument; 1 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/2011.13377.json · badge: /api/ran/2011.13377.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
linusericsson/ssl-transfer canonical (classifier label)
classifier basis: classifier_2025
4 ran · 1 no recorded run
FunctionStatusWhere it lives
count_acc Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
linusericsson/ssl-transfer/finetune.py
code served (permissive licence) · get_code("e628a3f24dbdf568")
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 float64, 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
from sklearn.metrics import confusion_matrix, precision_recall_curve

torch.manual_seed(0)

# Test case for 'accuracy' metric
pred_accuracy = torch.tensor([0, 1, 2, 1, 0, 2])
label_accuracy = torch.tensor([0, 1, 2, 2, 0, 1])

# Test case for 'mean per-class accuracy' metric
pred_mean_acc = torch.tensor([0, 1, 2, 1, 0, 2])
label_mean_acc = torch.tensor([0, 1, 2, 2, 0, 1])

# Test case for 'mAP' metric (multi-label classification)
# pred and label should be 2D tensors with shape (batch_size, num_classes)
pred_map = torch.tensor([
    [0.9, 0.1, 0.2, 0.8],
    [0.3, 0.7, 0.6, 0.2],
    [0.1, 0.8, 0.9, 0.3],
    [0.6, 0.2, 0.1, 0.7],
    [0.8, 0.3, 0.4, 0.5],
    [0.2, 0.9, 0.7, 0.1]
], dtype=torch.float32)

label_map = torch.tensor([
    [1, 0, 0, 1],
    [0, 1, 1, 0],
    [0, 1, 1, 0],
    [1, 0, 0, 1],
    [1, 0, 0, 1],
    [0, 1, 1, 0]
], dtype=torch.float32)

# Use the 'mAP' metric case
ARGS = {
    'pred': pred_map,
    'label': label_map,
    'metric': 'mAP'
}
Setup block drafted by
a language model, Claude Sonnet 4.5, prompt python_fixture_v1
Imports
numpy, sklearn, torch — 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
euclidean_dist Ran
ran; our contract draft was wrong, not the code
linusericsson/ssl-transfer/few_shot.py
code served (permissive licence) · get_code("4dd319c45d372246")
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 2, dtype family float, shape (2, 3) 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 two tensors with different numbers of samples but same feature dimension
# x: n samples with d features
# y: m samples with d features
# Using distinct sizes: n=2, m=3, d=4

x = torch.randn(2, 4)  # 2 samples, 4 features
y = torch.randn(3, 4)  # 3 samples, 4 features

ARGS = {
    'x': x,
    'y': y
}
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 2, dtype family float, shape (2, 3) — shape (2, 3) 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
voc_ap Ran
ran, honoured the contract we drafted
linusericsson/ssl-transfer/linear.py
code served (permissive licence) · get_code("fb52994d0ed2666d")
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 float64, 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 numpy as np

# Create sample recall and precision arrays
# These should be sorted arrays representing recall and precision values
# at different detection thresholds

# Typical scenario: as we lower the confidence threshold, recall increases
# but precision may decrease
rec = np.array([0.0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0])
prec = np.array([1.0, 0.95, 0.9, 0.85, 0.8, 0.75, 0.7, 0.65, 0.6, 0.55, 0.5])

ARGS = {
    'rec': rec,
    'prec': prec
}
Setup block drafted by
a language model, Claude Sonnet 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
voc_eval_cls Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
linusericsson/ssl-transfer/linear.py
code served (permissive licence) · get_code("da6e06cd5690114a")
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 float64, 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 numpy as np
from sklearn.metrics import precision_recall_curve

def voc_ap(rec, prec):
    ap = 0.0
    for t in np.linspace(0, 1, 11):
        if np.sum(rec >= t) == 0:
            p = 0
        else:
            p = np.max(prec[rec >= t])
        ap += p / 11.0
    return ap

def voc_eval_cls(y_true, y_pred):
    prec, rec, _ = precision_recall_curve(y_true, y_pred)
    ap = voc_ap(rec, prec)
    return ap

# Create binary classification ground truth and predictions
y_true = np.array([0, 0, 1, 1, 0, 1, 1, 0, 1, 0])
y_pred = np.array([0.1, 0.3, 0.8, 0.6, 0.2, 0.9, 0.7, 0.4, 0.85, 0.15])

ARGS = {
    'y_true': y_true,
    'y_pred': y_pred
}
Setup block drafted by
a language model, Claude Sonnet 4.5, prompt python_fixture_v1
Imports
numpy, sklearn — 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
get_dataset Not yet run
unverified (harvested, no recorded run)
linusericsson/ssl-transfer/linear.py
code served (permissive licence) · get_code("9f379d81ff970e3a")
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, 4 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 4 later) and 1 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
euclidean_dist 16 other papers
5 dated earlier, 11 later · 2 only as an assembly brick
1703.05175 (2017-03)
Prototypical Networks for Few-shot Learning
1902.08605 (2019-02)
Are Few-Shot Learning Benchmarks too Simple ? Solving them without Task Supervision at Test-Time
1905.13613 (2019-05)
Regression Networks for Meta-Learning Few-Shot Classification
and 13 more not named here
voc_ap 3 other papers
3 later
2308.15074 (2023-08)
Exploring Model Transferability through the Lens of Potential Energy
2312.17742 (2023-12)
Learning Vision from Models Rivals Learning Vision from Data
2502.16471 (2025-02)
Feature Space Perturbation: A Panacea to Enhanced Transferability Estimation
count_acc 2 other papers
2 later
2308.15074 (2023-08)
Exploring Model Transferability through the Lens of Potential Energy
2502.16471 (2025-02)
Feature Space Perturbation: A Panacea to Enhanced Transferability Estimation
voc_eval_cls 2 other papers
2 later
2308.15074 (2023-08)
Exploring Model Transferability through the Lens of Potential Energy
2502.16471 (2025-02)
Feature Space Perturbation: A Panacea to Enhanced Transferability Estimation

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

Self-supervised visual representation learning has seen huge progress recently, but no large scale evaluation has compared the many models now available. We evaluate the transfer performance of 13 top self-supervised models on 40 downstream tasks, including many-shot and few-shot recognition, object detection, and dense prediction. We compare their performance to a supervised baseline and show that on most tasks the best self-supervised models outperform supervision, confirming the recently observed trend in the literature. We find ImageNet Top-1 accuracy to be highly correlated with transfer to many-shot recognition, but increasingly less so for few-shot, object detection and dense prediction. No single self-supervised method dominates overall, suggesting that universal pre-training is still unsolved. Our analysis of features suggests that top self-supervised learners fail to preserve colour information as well as supervised alternatives, but tend to induce better classifier calibration, and less attentive overfitting than supervised learners.

For agents

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

get_harvested_code_for_paper("2011.13377")
get_code_for_paper("2011.13377")
have("2011.13377")

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

curl https://syntology.ai/api/ran/2011.13377.json

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

[![Syntology run record](https://syntology.ai/api/ran/2011.13377.svg)](https://syntology.ai/paper/2011.13377)

Connect an agent — have() is free.