SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1905.00561 · 2019

Large-scale weakly-supervised pre-training for video action recognition

arXiv · PDF · Open in the Atlas

Code that ran

12 of 21 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 11 have no contract verdict; 1 failure of our instrument; 9 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/1905.00561.json · badge: /api/ran/1905.00561.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
microsoft/computervision-recipes community (classifier label)
classifier basis not recorded on the link
8 ran · 8 no recorded run
moabitcoin/ig65m-pytorch community (classifier label)
classifier basis not recorded on the link
4 ran · 1 no recorded run
FunctionStatusWhere it lives
blobs_from_pkl Ran
ran on a synthesized input
moabitcoin/ig65m-pytorch/ig65m/cli/convert.py
code served (permissive licence) · get_code("8439277e085dfc38")
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 list, ndim 1, 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 pickle
import tempfile
from pathlib import Path

# Create a temporary pickle file with the expected structure
temp_dir = tempfile.mkdtemp()
temp_path = Path(temp_dir) / "test.pkl"

# Create sample blobs data
blobs_data = {
    'blobs': [
        {'id': 0, 'values': [1.0, 2.0, 3.0]},
        {'id': 1, 'values': [4.0, 5.0, 6.0]},
    ]
}

# Write the pickle file
with temp_path.open(mode='wb') as f:
    pickle.dump(blobs_data, f)

ARGS = {
    'path': temp_path
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
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
convert2gray Ran
ran on a synthesized input
microsoft/computervision-recipes/contrib/document_cleanup/light_weight_document_cleanup_ICDAR2021/model.py
code served (permissive licence) · get_code("2ae3a0bf00128d4b")
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 int, shape (2, 4, 5, 1) 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

# Create a small RGB image tensor
# Shape: (batch_size, height, width, channels)
# For RGB input, channels must be 3
batch_size = 2
height = 4
width = 5
channels = 3

# Create a random RGB tensor with values in valid image range [0, 255]
rgb_tensor = tf.constant(
    np.random.randint(0, 256, size=(batch_size, height, width, channels), dtype=np.uint8),
    dtype=tf.uint8
)

ARGS = {
    "in_tensor": rgb_tensor
}
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 int, shape (2, 4, 5, 1) — shape (2, 4, 5, 1) 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_optimal_threshold Ran
ran on a synthesized input
microsoft/computervision-recipes/utils_cv/classification/model.py
code served (permissive licence) · get_code("e057f034f658dbfe")
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 float64, ndim 0, 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 torch
import numpy as np
from typing import Callable, List
from torch import Tensor

torch.manual_seed(0)

# Define a simple metric function that takes y_pred, y_true, and threshold
def dummy_metric_function(y_pred: Tensor, y_true: Tensor, threshold: float) -> Tensor:
    """
    Simple metric function that computes accuracy based on threshold.
    Returns a scalar tensor.
    """
    predictions = (y_pred > threshold).float()
    accuracy = (predictions == y_true).float().mean()
    return accuracy

# Create small tensors for y_pred and y_true
# y_pred: predicted probabilities, shape (batch_size, num_classes) or (batch_size,)
# For binary classification: shape (2,) with values in [0, 1]
y_pred = torch.tensor([0.2, 0.8], dtype=torch.float32)
y_true = torch.tensor([0.0, 1.0], dtype=torch.float32)

# Create threshold list
thresholds = list(np.linspace(0, 1, 21))

ARGS = {
    "metric_function": dummy_metric_function,
    "y_pred": y_pred,
    "y_true": y_true,
    "thresholds": thresholds,
}
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 float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, 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
get_ratio_correct_metric Ran
ran on a synthesized input
microsoft/computervision-recipes/utils_cv/segmentation/model.py
code served (permissive licence) · get_code("f809ba52b7fe6bf7")
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 partial. 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

torch.manual_seed(0)

ARGS = {
    "classes": ["void", "person", "car", "bicycle"]
}
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 partial — a partial output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a partial 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
gram_matrix Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
microsoft/computervision-recipes/contrib/document_cleanup/light_weight_document_cleanup_ICDAR2021/loss_function.py
code served (permissive licence) · get_code("24caeadc560dd7b1")
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 3, dtype family float, shape (2, 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 tensorflow as tf
import numpy as np

# Create a small input tensor for gram_matrix
# The function expects shape (batch, height, width, channels)
# Using distinct dimensions: batch=2, height=4, width=3, channels=5
batch_size = 2
height = 4
width = 3
channels = 5

input_tensor = tf.constant(
    np.random.randn(batch_size, height, width, channels).astype(np.float32)
)

ARGS = {
    "input_tensor": input_tensor
}
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 3, dtype family float, shape (2, 5, 5) — shape (2, 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
hamming_accuracy Ran
ran on a synthesized input
microsoft/computervision-recipes/utils_cv/classification/model.py
code served (permissive licence) · get_code("6e3c07a22fc403cd")
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
from torch import Tensor

torch.manual_seed(0)

# Create small tensors for testing
# y_pred: raw predictions (before sigmoid), shape [2, 4]
y_pred = torch.randn(2, 4)

# y_true: binary labels, shape [2, 4]
y_true = torch.randint(0, 2, (2, 4)).float()

# threshold: float value
threshold = 0.2

# sigmoid: boolean flag
sigmoid = False

ARGS = {
    "y_pred": y_pred,
    "y_true": y_true,
    "threshold": threshold,
    "sigmoid": sigmoid,
}
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
is_installed Ran
ran on a synthesized input
microsoft/computervision-recipes/contrib/vm_builder/vm_builder.py
code served (permissive licence) · get_code("8baa827e7a093444")
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 bool, 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 shutil

# The function checks if a CLI application is on PATH and executable
# It uses shutil.which() which takes a string command name
# We need to provide a cli_app string that represents a common executable

# Using a standard Unix/Linux/Windows utility that should be available
# Common options: 'python', 'ls', 'cat', 'echo', 'git', 'curl', etc.
# 'python' is most portable since we're running Python code

ARGS = {
    "cli_app": "python"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, 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
r2plus1d_34_32_ig65m Ran
ran on a synthesized input
moabitcoin/ig65m-pytorch/ig65m/models.py
code served (permissive licence) · get_code("cb39d2a70fde7d87")
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 VideoResNet. 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
from torchvision.models.video.resnet import VideoResNet, BasicBlock, R2Plus1dStem, Conv2Plus1D

torch.manual_seed(0)

ARGS = {
    'num_classes': 359,
    'pretrained': False,
    'progress': False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, torchvision — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type VideoResNet — a VideoResNet output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a VideoResNet 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
r2plus1d_34_8_ig65m Ran
ran on a synthesized input
moabitcoin/ig65m-pytorch/ig65m/models.py
code served (permissive licence) · get_code("21dfb37daa3d5a63")
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 VideoResNet. 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
from torchvision.models.video.resnet import VideoResNet, BasicBlock, R2Plus1dStem, Conv2Plus1D

torch.manual_seed(0)

ARGS = {
    'num_classes': 487,
    'pretrained': False,
    'progress': False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, torchvision — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type VideoResNet — a VideoResNet output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a VideoResNet 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
r2plus1d_34_8_kinetics Ran
ran on a synthesized input
moabitcoin/ig65m-pytorch/ig65m/models.py
code served (permissive licence) · get_code("ba55356a87900992")
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 VideoResNet. 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
from torchvision.models.video.resnet import VideoResNet, BasicBlock, R2Plus1dStem, Conv2Plus1D

torch.manual_seed(0)

ARGS = {
    'num_classes': 400,
    'pretrained': False,
    'progress': False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, torchvision — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type VideoResNet — a VideoResNet output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a VideoResNet 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
ratio_correct Ran
ran on a synthesized input
microsoft/computervision-recipes/utils_cv/segmentation/model.py
code served (permissive licence) · get_code("c9d0ef61d7a0052a")
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)

# The function expects:
# - void_id: an integer or None
# - input: a tensor with shape (batch_size, num_classes, height, width) 
#          that will be argmax'd on dim=1
# - target: a tensor with shape (batch_size, 1, height, width) 
#           that will be squeezed on dim=1

# Create small tensors
batch_size = 2
num_classes = 4
height = 3
width = 3

# input tensor: (batch, num_classes, height, width)
# After argmax(dim=1), it becomes (batch, height, width)
input_tensor = torch.randn(batch_size, num_classes, height, width)

# target tensor: (batch, 1, height, width)
# After squeeze(1), it becomes (batch, height, width)
target_tensor = torch.randint(0, num_classes, (batch_size, 1, height, width))

# void_id can be None or an integer in the range of class indices
void_id = 0

ARGS = {
    "void_id": void_id,
    "input": input_tensor,
    "target": target_tensor
}
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
zero_one_accuracy Ran
ran on a synthesized input
microsoft/computervision-recipes/utils_cv/classification/model.py
code served (permissive licence) · get_code("19b949c8e8b6cc5a")
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
from torch import Tensor

torch.manual_seed(0)

# Create small prediction and true label tensors
# batch size: 2, number of labels: 4
y_pred = torch.randn(2, 4)  # predictions before sigmoid/threshold
y_true = torch.randint(0, 2, (2, 4)).float()  # binary true labels

# Define ARGS dict with all parameters
ARGS = {
    "y_pred": y_pred,
    "y_true": y_true,
    "threshold": 0.2,
    "sigmoid": 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 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
Compute_PLoss Not yet run
unverified (harvested, no recorded run)
microsoft/computervision-recipes/contrib/document_cleanup/light_weight_document_cleanup_ICDAR2021/loss_function.py
code served (permissive licence) · get_code("b79851f2da94bb38")
How it ran

no invocation recorded.

CreateModel_M16_binary Not yet run
unverified (harvested, no recorded run)
microsoft/computervision-recipes/contrib/document_cleanup/light_weight_document_cleanup_ICDAR2021/model.py
code served (permissive licence) · get_code("d2e1797dd9a3e283")
How it ran

no invocation recorded.

evaluate_mot Not yet run
unverified (harvested, no recorded run)
microsoft/computervision-recipes/utils_cv/tracking/model.py
code served (permissive licence) · get_code("ca5129f92643bc3e")
How it ran

no invocation recorded.

get_pretrained_fasterrcnn Not yet run
unverified (harvested, no recorded run)
microsoft/computervision-recipes/utils_cv/detection/model.py
code served (permissive licence) · get_code("6f8cd1e7f213cb4a")
How it ran

no invocation recorded.

get_pretrained_keypointrcnn Not yet run
unverified (harvested, no recorded run)
microsoft/computervision-recipes/utils_cv/detection/model.py
code served (permissive licence) · get_code("f763516cdbd12500")
How it ran

no invocation recorded.

get_pretrained_maskrcnn Not yet run
unverified (harvested, no recorded run)
microsoft/computervision-recipes/utils_cv/detection/model.py
code served (permissive licence) · get_code("3769b815f2bf3b26")
How it ran

no invocation recorded.

main Not yet run
unverified (harvested, no recorded run)
moabitcoin/ig65m-pytorch/ig65m/cli/server.py
code served (permissive licence) · get_code("1a80caf172692d16")
How it ran

no invocation recorded.

res_net_block Not yet run
unverified (harvested, no recorded run)
microsoft/computervision-recipes/contrib/document_cleanup/light_weight_document_cleanup_ICDAR2021/model.py
code served (permissive licence) · get_code("6da0ad792057efe4")
How it ran

no invocation recorded.

vgg_layers Not yet run
unverified (harvested, no recorded run)
microsoft/computervision-recipes/contrib/document_cleanup/light_weight_document_cleanup_ICDAR2021/loss_function.py
code served (permissive licence) · get_code("c9e8ae02d557b195")
How it ran

no invocation recorded.

Also present in other papers' code

Of 21 samples lifted from this paper's linked repositories, 21 carry a function identity and 0 do not. Of the 21 distinct functions with an identity, 2 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 0 the same month, 2 later) and 19 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
gram_matrix 2 other papers
1 dated earlier, 1 later
1508.06576 (2015-08)
A Neural Algorithm of Artistic Style
2108.07379 (2021-08)
DRB-GAN: A Dynamic ResBlock Generative Adversarial Network for Artistic Style Transfer
vgg_layers 2 other papers
1 dated earlier, 1 later
1508.06576 (2015-08)
A Neural Algorithm of Artistic Style
2108.07379 (2021-08)
DRB-GAN: A Dynamic ResBlock Generative Adversarial Network for Artistic Style Transfer

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

Current fully-supervised video datasets consist of only a few hundred thousand videos and fewer than a thousand domain-specific labels. This hinders the progress towards advanced video architectures. This paper presents an in-depth study of using large volumes of web videos for pre-training video models for the task of action recognition. Our primary empirical finding is that pre-training at a very large scale (over 65 million videos), despite on noisy social-media videos and hashtags, substantially improves the state-of-the-art on three challenging public action recognition datasets. Further, we examine three questions in the construction of weakly-supervised video action datasets. First, given that actions involve interactions with objects, how should one construct a verb-object pre-training label space to benefit transfer learning the most? Second, frame-based models perform quite well on action recognition; is pre-training for good image features sufficient or is pre-training for spatio-temporal features valuable for optimal transfer learning? Finally, actions are generally less well-localized in long videos vs. short videos; since action labels are provided at a video level, how should one choose video clips for best performance, given some fixed budget of number or minutes of videos?

For agents

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

get_harvested_code_for_paper("1905.00561")
get_code_for_paper("1905.00561")
have("1905.00561")

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

curl https://syntology.ai/api/ran/1905.00561.json

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

[![Syntology run record](https://syntology.ai/api/ran/1905.00561.svg)](https://syntology.ai/paper/1905.00561)

Connect an agent — have() is free.