SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1703.04044 · 2017

Colorization as a Proxy Task for Visual Understanding

arXiv · PDF · Open in the Atlas

Code that ran

5 of 10 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 5 have no contract verdict; 0 failures of our instrument; 5 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/1703.04044.json · badge: /api/ran/1703.04044.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
gustavla/self-supervision canonical (classifier label)
classifier basis: census
5 ran · 5 no recorded run
FunctionStatusWhere it lives
conv Ran
ran on a synthesized input
gustavla/self-supervision/selfsup/jigsaw.py
code served (permissive licence) · get_code("a9966f2e69338f0a")
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 EagerTensor, ndim 4, dtype family float, shape (2, 2, 2, 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 tensorflow as tf
import numpy as np

# Set seed for reproducibility
tf.random.set_seed(0)

# Create a small batch of input data
# Shape: [batch_size, height, width, channels_in]
batch_size = 2
height = 4
width = 4
c_i = 3  # input channels
input_tensor = tf.constant(np.random.randn(batch_size, height, width, c_i).astype(np.float32))

# Create kernel (filter)
# Shape: [kernel_height, kernel_width, channels_in, channels_out]
k_h = 3
k_w = 3
c_o = 5  # output channels
kernel = tf.constant(np.random.randn(k_h, k_w, c_i, c_o).astype(np.float32))

# Create biases
# Shape: [channels_out]
biases = tf.constant(np.random.randn(c_o).astype(np.float32))

# Stride parameters
s_h = 1
s_w = 1

# Build the arguments dictionary
ARGS = {
    'input': input_tensor,
    'kernel': kernel,
    'biases': biases,
    'k_h': k_h,
    'k_w': k_w,
    'c_o': c_o,
    's_h': s_h,
    's_w': s_w,
    'padding': 'VALID',
    'group': 1,
    'name': 'test_conv'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, ndim 4, dtype family float, shape (2, 2, 2, 5) — shape (2, 2, 2, 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
conv Ran
ran on a synthesized input
gustavla/self-supervision/selfsup/model/alex.py
code served (permissive licence) · get_code("51b2d008e6210eb5")
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 EagerTensor, ndim 4, dtype family float, shape (2, 2, 2, 8) 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 tensorflow as tf
import numpy as np

# Set seed for reproducibility
tf.random.set_seed(0)
np.random.seed(0)

# Create small tensors for testing
# input: [batch_size, height, width, channels_in]
batch_size = 2
height = 4
width = 4
channels_in = 3
channels_out = 8

input_tensor = tf.random.normal([batch_size, height, width, channels_in])

# kernel: [kernel_height, kernel_width, channels_in, channels_out]
k_h = 3
k_w = 3
kernel = tf.random.normal([k_h, k_w, channels_in, channels_out])

# biases: [channels_out]
biases = tf.random.normal([channels_out])

# stride values
s_h = 1
s_w = 1

# output channels
c_o = channels_out

# padding type
padding = 'VALID'

# group parameter (1 means standard convolution)
group = 1

ARGS = {
    'input': input_tensor,
    'kernel': kernel,
    'biases': biases,
    'k_h': k_h,
    'k_w': k_w,
    'c_o': c_o,
    's_h': s_h,
    's_w': s_w,
    'padding': padding,
    'group': group
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, ndim 4, dtype family float, shape (2, 2, 2, 8) — shape (2, 2, 2, 8) 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
create Ran
ran on a synthesized input
gustavla/self-supervision/selfsup/info.py
code served (permissive licence) · get_code("8359957dc18b9b75")
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 dict. 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
from collections import OrderedDict

torch.manual_seed(0)

ARGS = {
    'scale_summary': 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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
leaky_relu Ran
ran on a synthesized input
gustavla/self-supervision/selfsup/model/vgg16_bn.py
code served (permissive licence) · get_code("e9208491625e9309")
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 EagerTensor, ndim 2, 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 tensorflow as tf
import numpy as np

def leaky_relu(x):
    return tf.maximum(0.01 * x, x)

# Create a small TensorFlow tensor
x = tf.constant(np.array([[-2.0, -1.0, 0.5, 1.5],
                           [3.0, -0.5, 2.0, -3.0]], dtype=np.float32))

ARGS = {
    'x': x
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, ndim 2, dtype family float — no output shape is stated: observe_signature.py recorded SymbolicTensor ndim 2, which disagrees with the fixture run
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
load_permutations Ran
ran on a synthesized input
gustavla/self-supervision/selfsup/jigsaw.py
code served (permissive licence) · get_code("4ca39f8eea1448b8")
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 int, shape (3, 9) 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
import struct
import tempfile
import os

# Create a temporary binary file with valid permutation data
temp_fd, temp_path = tempfile.mkstemp(suffix='.bin')
try:
    with os.fdopen(temp_fd, 'wb') as f:
        N = 3  # Number of permutations
        P = 9  # Size of each permutation
        
        # Write header
        f.write(struct.pack('<II', N, P))
        
        # Write permutations
        for i in range(N):
            perm = tuple(range(P))
            f.write(struct.pack('<IIIIIIIII', *perm))
    
    ARGS = {
        'fn': temp_path
    }
finally:
    pass
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 int, shape (3, 9) — shape (3, 9) 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
decode_image Not yet run
unverified (harvested, no recorded run)
gustavla/self-supervision/selfsup/datasets.py
code served (permissive licence) · get_code("00c3824b77b7608e")
How it ran

no invocation recorded.

init_batch_norm_vars Not yet run
unverified (harvested, no recorded run)
gustavla/self-supervision/selfsup/model/alex_bn.py
code served (permissive licence) · get_code("37b8cb185d7de432")
How it ran

no invocation recorded.

moments Not yet run
unverified (harvested, no recorded run)
gustavla/self-supervision/selfsup/extra.py
code served (permissive licence) · get_code("26acd189772e5c98")
How it ran

no invocation recorded.

normalize_moments Not yet run
unverified (harvested, no recorded run)
gustavla/self-supervision/selfsup/extra.py
code served (permissive licence) · get_code("0b335fc675462059")
How it ran

no invocation recorded.

resize_to_ensure_size Not yet run
unverified (harvested, no recorded run)
gustavla/self-supervision/selfsup/datasets.py
code served (permissive licence) · get_code("58bb5de8eea4c169")
How it ran

no invocation recorded.

Also present in other papers' code

Of 10 samples lifted from this paper's linked repositories, 10 carry a function identity and 0 do not. Of the 10 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 10 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.

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

We investigate and improve self-supervision as a drop-in replacement for ImageNet pretraining, focusing on automatic colorization as the proxy task. Self-supervised training has been shown to be more promising for utilizing unlabeled data than other, traditional unsupervised learning methods. We build on this success and evaluate the ability of our self-supervised network in several contexts. On VOC segmentation and classification tasks, we present results that are state-of-the-art among methods not using ImageNet labels for pretraining representations. Moreover, we present the first in-depth analysis of self-supervision via colorization, concluding that formulation of the loss, training details and network architecture play important roles in its effectiveness. This investigation is further expanded by revisiting the ImageNet pretraining paradigm, asking questions such as: How much training data is needed? How many labels are needed? How much do features change when fine-tuned? We relate these questions back to self-supervision by showing that colorization provides a similarly powerful supervisory signal as various flavors of ImageNet pretraining.

For agents

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

get_harvested_code_for_paper("1703.04044")
get_code_for_paper("1703.04044")
have("1703.04044")

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

curl https://syntology.ai/api/ran/1703.04044.json

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

[![Syntology run record](https://syntology.ai/api/ran/1703.04044.svg)](https://syntology.ai/paper/1703.04044)

Connect an agent — have() is free.