SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2402.13616 · ECCV · 2024

YOLOv9: Learning What You Want to Learn Using Programmable Gradient Information

Chien-Yao Wang, I-Hau Yeh, Hong-Yuan Liao

arXiv · PDF · Open in the Atlas

Code that ran

25 of 32 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (9 of them constructed an object rather than computing a result; 5 honoured the contract we drafted, 2 violated it and 13 have no contract verdict; 5 failures of our instrument; 7 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/2402.13616.json · badge: /api/ran/2402.13616.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
WongKinYiu/YOLO canonical (classifier label)
classifier basis: census
7 ran · 2 no recorded run
WongKinYiu/yolov9 canonical (classifier label)
classifier basis not recorded on the link
2 ran · 1 no recorded run
henrytsui000/YOLO community
Papers with Code archive
7 ran · 2 no recorded run
ultralytics/ultralytics named in the paper
Papers with Code archive
6 ran · 2 no recorded run
ibaiGorordo/ONNX-YOLOv9-MIT-Object-Detection community (classifier label)
classifier basis not recorded on the link
3 ran · 0 no recorded run
FunctionStatusWhere it lives
Bottleneck Ran
ran on a synthesized input · constructed an object; no result was computed
ultralytics/ultralytics/ultralytics/nn/modules/block.py
pointer only (licence: AGPL-3.0) · get_code("7707093cd2ca2b68")
How it ran

The body of this sample is withheld: AGPL-3.0, as GitHub detects it on this paper's repository, ultralytics/ultralytics, is not in the permissive set Syntology serves inline. 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.

Bottleneck Ran
ran on a synthesized input · constructed an object; no result was computed
henrytsui000/YOLO/yolo/model/module.py
code served (permissive licence) · get_code("dea6d7ba4c8c882f")
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, Bottleneck: the run constructed the object and did not call it, so no result was computed. 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 typing import Tuple

torch.manual_seed(0)

# Helper functions from the code
def auto_pad(kernel_size, dilation=1, **kwargs) -> Tuple[int, int]:
    """Auto Padding for the convolution blocks"""
    if isinstance(kernel_size, int):
        kernel_size = (kernel_size, kernel_size)
    if isinstance(dilation, int):
        dilation = (dilation, dilation)
    pad_h = (kernel_size[0] - 1) * dilation[0] // 2
    pad_w = (kernel_size[1] - 1) * dilation[1] // 2
    return (pad_h, pad_w)

def create_activation_function(activation: str) -> nn.Module:
    """Retrieves an activation function from the PyTorch nn module based on its name, case-insensitively."""
    if not activation or activation.lower() in ['false', 'none']:
        return nn.Identity()
    activation_map = {name.lower(): obj for name, obj in nn.modules.activation.__dict__.items() if isinstance(obj, type) and issubclass(obj, nn.Module)}
    if activation.lower() in activation_map:
        return activation_map[activation.lower()](inplace=True)
    else:
        raise ValueError(f"Activation function '{activation}' is not found in torch.nn")

class Conv(nn.Module):
    """A basic convolutional block that includes convolution, batch normalization, and activation."""

    def __init__(self, in_channels: int, out_channels: int, kernel_size, *, activation: str='SiLU', **kwargs):
        super().__init__()
        kwargs.setdefault('padding', auto_pad(kernel_size, **kwargs))
        self.conv = nn.Conv2d(in_channels, out_channels, kernel_size, bias=False, **kwargs)
        self.bn = nn.BatchNorm2d(out_channels, eps=0.001, momentum=0.03)
        self.act = create_activation_function(activation)

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        return self.act(self.bn(self.conv(x)))

class RepConv(nn.Module):
    """A convolutional block that combines two convolution layers (kernel and point-wise)."""

    def __init__(self, in_channels: int, out_channels: int, kernel_size=3, *, activation: str='SiLU', **kwargs):
        super().__init__()
        self.act = create_activation_function(activation)
        self.conv1 = Conv(in_channels, out_channels, kernel_size, activation=False, **kwargs)
        self.conv2 = Conv(in_channels, out_channels, 1, activation=False, **kwargs)

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        return self.act(self.conv1(x) + self.conv2(x))

class Bottleneck(nn.Module):
    """A bottleneck block with optional residual connections."""

    def __init__(self, in_channels: int, out_channels: int, *, kernel_size: Tuple[int, int]=(3, 3), residual: bool=True, expand: float=1.0, **kwargs):
        super().__init__()
        neck_channels = int(out_channels * expand)
        self.conv1 = RepConv(in_channels, neck_channels, kernel_size[0], **kwargs)
        self.conv2 = Conv(neck_channels, out_channels, kernel_size[1], **kwargs)
        self.residual = residual
        if residual and in_channels != out_channels:
            self.residual = False

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        y = self.conv2(self.conv1(x))
        return x + y if self.residual else y

# Create test input tensor: batch=2, channels=4, height=8, width=8
test_input = torch.randn(2, 4, 8, 8)

ARGS = {
    'in_channels': 4,
    'out_channels': 8,
    'kernel_size': (3, 3),
    'residual': True,
    'expand': 1.0,
}
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 Bottleneck — 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 Bottleneck 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
C3 Ran
ran on a synthesized input · constructed an object; no result was computed
ultralytics/ultralytics/ultralytics/nn/modules/block.py
pointer only (licence: AGPL-3.0) · get_code("955083ecf0b31e55")
How it ran

The body of this sample is withheld: AGPL-3.0, as GitHub detects it on this paper's repository, ultralytics/ultralytics, is not in the permissive set Syntology serves inline. 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.

Conv Ran
ran on a synthesized input · constructed an object; no result was computed
ultralytics/ultralytics/ultralytics/nn/modules/block.py
pointer only (licence: AGPL-3.0) · get_code("af6c4f0d1c425989")
How it ran

The body of this sample is withheld: AGPL-3.0, as GitHub detects it on this paper's repository, ultralytics/ultralytics, is not in the permissive set Syntology serves inline. 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.

Conv Ran
ran on a synthesized input · constructed an object; no result was computed
henrytsui000/YOLO/yolo/model/module.py
code served (permissive licence) · get_code("f12c179ff392638b")
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, Conv: the run constructed the object and did not call it, so no result was computed. 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 torch.nn.common_types import _size_2_t

torch.manual_seed(0)

ARGS = {
    'in_channels': 3,
    'out_channels': 16,
    'kernel_size': 3,
    'activation': 'SiLU'
}
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 Conv — 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 Conv 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
DConv Ran
ran on a synthesized input · constructed an object; no result was computed
henrytsui000/YOLO/yolo/model/module.py
code served (permissive licence) · get_code("859cd8d4d3b28b71")
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, DConv: the run constructed the object and did not call it, so no result was computed. 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)

ARGS = {
    'in_channels': 512,
    'alpha': 0.8,
    'atoms': 512,
}
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 DConv — 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 DConv 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
RepBottleneck Ran
ran on a synthesized input · constructed an object; no result was computed
ultralytics/ultralytics/ultralytics/nn/modules/block.py
pointer only (licence: AGPL-3.0) · get_code("7385e2e3d2a2af23")
How it ran

The body of this sample is withheld: AGPL-3.0, as GitHub detects it on this paper's repository, ultralytics/ultralytics, is not in the permissive set Syntology serves inline. 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.

RepConv Ran
ran on a synthesized input · constructed an object; no result was computed
ultralytics/ultralytics/ultralytics/nn/modules/block.py
pointer only (licence: AGPL-3.0) · get_code("a36954bb2b2d92af")
How it ran

The body of this sample is withheld: AGPL-3.0, as GitHub detects it on this paper's repository, ultralytics/ultralytics, is not in the permissive set Syntology serves inline. 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.

RepConv Ran
ran on a synthesized input · constructed an object; no result was computed
henrytsui000/YOLO/yolo/model/module.py
code served (permissive licence) · get_code("5ddf9e8137680cff")
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, RepConv: the run constructed the object and did not call it, so no result was computed. 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 torch.nn.common_types import _size_2_t

torch.manual_seed(0)

def auto_pad(kernel_size: _size_2_t, dilation: _size_2_t=1, **kwargs) -> tuple:
    """
    Auto Padding for the convolution blocks
    """
    if isinstance(kernel_size, int):
        kernel_size = (kernel_size, kernel_size)
    if isinstance(dilation, int):
        dilation = (dilation, dilation)
    pad_h = (kernel_size[0] - 1) * dilation[0] // 2
    pad_w = (kernel_size[1] - 1) * dilation[1] // 2
    return (pad_h, pad_w)

def create_activation_function(activation: str) -> nn.Module:
    """
    Retrieves an activation function from the PyTorch nn module based on its name, case-insensitively.
    """
    if not activation or activation.lower() in ['false', 'none']:
        return nn.Identity()
    activation_map = {name.lower(): obj for name, obj in nn.modules.activation.__dict__.items() if isinstance(obj, type) and issubclass(obj, nn.Module)}
    if activation.lower() in activation_map:
        return activation_map[activation.lower()](inplace=True)
    else:
        raise ValueError(f"Activation function '{activation}' is not found in torch.nn")

class Conv(nn.Module):
    """A basic convolutional block that includes convolution, batch normalization, and activation."""

    def __init__(self, in_channels: int, out_channels: int, kernel_size: _size_2_t, *, activation=None, **kwargs):
        super().__init__()
        kwargs.setdefault('padding', auto_pad(kernel_size, **kwargs))
        self.conv = nn.Conv2d(in_channels, out_channels, kernel_size, bias=False, **kwargs)
        self.bn = nn.BatchNorm2d(out_channels, eps=0.001, momentum=0.03)
        self.act = create_activation_function(activation) if activation else nn.Identity()

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        return self.act(self.bn(self.conv(x)))

class RepConv(nn.Module):
    """A convolutional block that combines two convolution layers (kernel and point-wise)."""

    def __init__(self, in_channels: int, out_channels: int, kernel_size: _size_2_t=3, *, activation=None, **kwargs):
        super().__init__()
        self.act = create_activation_function(activation) if activation else nn.Identity()
        self.conv1 = Conv(in_channels, out_channels, kernel_size, activation=None, **kwargs)
        self.conv2 = Conv(in_channels, out_channels, 1, activation=None, **kwargs)

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        return self.act(self.conv1(x) + self.conv2(x))

ARGS = {
    'in_channels': 3,
    'out_channels': 4,
    'kernel_size': 3,
    'activation': 'SiLU',
}
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 RepConv — 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 RepConv 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
RepNCSP Ran
ran on a synthesized input
henrytsui000/YOLO/yolo/model/module.py
code served (permissive licence) · get_code("7115585ff5427d1d")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 4, 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 torch.nn as nn
from torch.nn.common_types import _size_2_t
import logging
from typing import Any, Dict, List, Optional, Tuple

logging.basicConfig(level=logging.WARNING)
logger = logging.getLogger('yolo')

def auto_pad(kernel_size: _size_2_t, dilation: _size_2_t=1, **kwargs) -> Tuple[int, int]:
    """
    Auto Padding for the convolution blocks
    """
    if isinstance(kernel_size, int):
        kernel_size = (kernel_size, kernel_size)
    if isinstance(dilation, int):
        dilation = (dilation, dilation)
    pad_h = (kernel_size[0] - 1) * dilation[0] // 2
    pad_w = (kernel_size[1] - 1) * dilation[1] // 2
    return (pad_h, pad_w)

def create_activation_function(activation: str) -> nn.Module:
    """
    Retrieves an activation function from the PyTorch nn module based on its name, case-insensitively.
    """
    if not activation or activation.lower() in ['false', 'none']:
        return nn.Identity()
    activation_map = {name.lower(): obj for name, obj in nn.modules.activation.__dict__.items() if isinstance(obj, type) and issubclass(obj, nn.Module)}
    if activation.lower() in activation_map:
        return activation_map[activation.lower()](inplace=True)
    else:
        raise ValueError(f"Activation function '{activation}' is not found in torch.nn")

class Conv(nn.Module):
    """A basic convolutional block that includes convolution, batch normalization, and activation."""

    def __init__(self, in_channels: int, out_channels: int, kernel_size: _size_2_t, *, activation: Optional[str]='SiLU', **kwargs):
        super().__init__()
        kwargs.setdefault('padding', auto_pad(kernel_size, **kwargs))
        self.conv = nn.Conv2d(in_channels, out_channels, kernel_size, bias=False, **kwargs)
        self.bn = nn.BatchNorm2d(out_channels, eps=0.001, momentum=0.03)
        self.act = create_activation_function(activation)

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        return self.act(self.bn(self.conv(x)))

class RepConv(nn.Module):
    """A convolutional block that combines two convolution layers (kernel and point-wise)."""

    def __init__(self, in_channels: int, out_channels: int, kernel_size: _size_2_t=3, *, activation: Optional[str]='SiLU', **kwargs):
        super().__init__()
        self.act = create_activation_function(activation)
        self.conv1 = Conv(in_channels, out_channels, kernel_size, activation=False, **kwargs)
        self.conv2 = Conv(in_channels, out_channels, 1, activation=False, **kwargs)

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        return self.act(self.conv1(x) + self.conv2(x))

class Bottleneck(nn.Module):
    """A bottleneck block with optional residual connections."""

    def __init__(self, in_channels: int, out_channels: int, *, kernel_size: Tuple[int, int]=(3, 3), residual: bool=True, expand: float=1.0, **kwargs):
        super().__init__()
        neck_channels = int(out_channels * expand)
        self.conv1 = RepConv(in_channels, neck_channels, kernel_size[0], **kwargs)
        self.conv2 = Conv(neck_channels, out_channels, kernel_size[1], **kwargs)
        self.residual = residual
        if residual and in_channels != out_channels:
            self.residual = False
            logger.warning('Residual connection disabled: in_channels (%d) != out_channels (%d)', in_channels, out_channels)

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        y = self.conv2(self.conv1(x))
        return x + y if self.residual else y

class RepNCSP(nn.Module):
    """RepNCSP block with convolutions, split, and bottleneck processing."""

    def __init__(self, in_channels: int, out_channels: int, kernel_size: int=1, *, csp_expand: float=0.5, repeat_num: int=1, neck_args: Dict[str, Any]={}, **kwargs):
        super().__init__()
        neck_channels = int(out_channels * csp_expand)
        self.conv1 = Conv(in_channels, neck_channels, kernel_size, **kwargs)
        self.conv2 = Conv(in_channels, neck_channels, kernel_size, **kwargs)
        self.conv3 = Conv(2 * neck_channels, out_channels, kernel_size, **kwargs)
        self.bottleneck = nn.Sequential(*[Bottleneck(neck_channels, neck_channels, **neck_args) for _ in range(repeat_num)])

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        x1 = self.bottleneck(self.conv1(x))
        x2 = self.conv2(x)
        return self.conv3(torch.cat((x1, x2), dim=1))

INIT_ARGS = {
    'in_channels': 4,
    'out_channels': 8,
    'kernel_size': 1,
    'csp_expand': 0.5,
    'repeat_num': 2,
    'neck_args': {},
}

ARGS = {
    'x': torch.randn(2, 4, 16, 16),
}

if __name__ == '__main__':
    model = RepNCSP(**INIT_ARGS)
    output = model(**ARGS)
    print(f"Output shape: {output.shape}")
    print(f"Expected shape: (2, 8, 16, 16)")
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
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 4, dtype family float — no output shape is stated: observe_signature.py observed this class with its constructor sizes canonicalised where it could, which the node does not record, so its shape is not stated as this setup block's
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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
auto_pad Ran
ran, violated the contract we drafted
henrytsui000/YOLO/yolo/model/module.py
code served (permissive licence) · get_code("00279341e44c9e1a")
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 tuple, ndim 1, dtype family int; 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 torch.nn as nn
from typing import Tuple

torch.manual_seed(0)

# The function auto_pad takes kernel_size and dilation as parameters
# kernel_size: _size_2_t which can be int or tuple of 2 ints
# dilation: _size_2_t which can be int or tuple of 2 ints (default=1)
# **kwargs: additional keyword arguments that are accepted but not used in the function body

# Test with tuple inputs
ARGS = {
    "kernel_size": (3, 5),
    "dilation": (1, 2),
}
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 tuple, ndim 1, dtype family int — a tuple 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
auto_pad Ran
ran, violated the contract we drafted
WongKinYiu/YOLO/yolo/utils/module_utils.py
code served (permissive licence) · get_code("acb86aa8696bb043")
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 tuple, ndim 1, dtype family int; 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
from torch.nn.common_types import _size_2_t

torch.manual_seed(0)

# The function auto_pad takes kernel_size and dilation as parameters
# kernel_size: _size_2_t can be int or tuple of 2 ints
# dilation: _size_2_t can be int or tuple of 2 ints, defaults to 1
# **kwargs: additional keyword arguments (not used in function body)

ARGS = {
    "kernel_size": (3, 5),
    "dilation": (1, 2),
}
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 tuple, ndim 1, dtype family int — a tuple 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
autopad Ran
ran, honoured the contract we drafted
ultralytics/ultralytics/ultralytics/nn/modules/block.py
pointer only (licence: AGPL-3.0) · get_code("11ea59c9308130b2")
How it ran

The body of this sample is withheld: AGPL-3.0, as GitHub detects it on this paper's repository, ultralytics/ultralytics, is not in the permissive set Syntology serves inline. 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.

autopad Ran
ran, honoured the contract we drafted
WongKinYiu/yolov9/models/common.py
pointer only (licence: GPL-3.0); the stored copy get_code serves, from LiuZhu-CV/DEAL, is pointer only (licence: NONE) · get_code("86db6b7ab1646a7c")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, WongKinYiu/yolov9, is not in the permissive set Syntology serves inline. 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.

calculate_iou Ran
ran; our contract draft was wrong, not the code
WongKinYiu/YOLO/yolo/utils/bounding_box_utils.py
code served (permissive licence) · get_code("c9cb18bff87ea33b")
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 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
import math

torch.manual_seed(0)

# Create bounding boxes in format [xmin, ymin, xmax, ymax]
# bbox1: 2 boxes
bbox1 = torch.tensor([
    [10.0, 20.0, 50.0, 60.0],
    [15.0, 25.0, 55.0, 65.0]
], dtype=torch.float32)

# bbox2: 3 boxes
bbox2 = torch.tensor([
    [12.0, 22.0, 52.0, 62.0],
    [20.0, 30.0, 60.0, 70.0],
    [5.0, 15.0, 45.0, 55.0]
], dtype=torch.float32)

# Test with different metrics
metrics = 'iou'

ARGS = {
    'bbox1': bbox1,
    'bbox2': bbox2,
    'metrics': metrics
}
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 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
create_activation_function Ran
ran; our contract draft was wrong, not the code
henrytsui000/YOLO/yolo/model/module.py
code served (permissive licence) · get_code("d0374cf3dbdab1f6")
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 ReLU. 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 function takes a single parameter: activation (a string)
# It retrieves activation functions from torch.nn by name, case-insensitively
# Valid activation functions include: ReLU, Sigmoid, Tanh, etc.

ARGS = {
    "activation": "ReLU"
}
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 ReLU — a ReLU output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a ReLU 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
draw_box Ran
ran on a synthesized input
ibaiGorordo/ONNX-YOLOv9-MIT-Object-Detection/yolov9/utils.py
code served (permissive licence) · get_code("b8d4eb0708db9812")
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 3, dtype family int, shape (100, 100, 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 cv2
import numpy as np

# Create a small test image (100x100 RGB image)
test_image = np.ones((100, 100, 3), dtype=np.uint8) * 255

# Create a bounding box [x1, y1, x2, y2]
test_box = np.array([10, 20, 50, 60], dtype=np.float32)

# Define the arguments dictionary
ARGS = {
    "image": test_image,
    "box": test_box,
    "color": (0, 0, 255),
    "thickness": 2
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
cv2, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family int, shape (100, 100, 3) — shape (100, 100, 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
draw_detections Ran
ran on a synthesized input
ibaiGorordo/ONNX-YOLOv9-MIT-Object-Detection/yolov9/utils.py
code served (permissive licence) · get_code("c9dab4a605d6462a")
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 3, dtype family int, shape (480, 640, 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 cv2
import numpy as np

class_names = ['person', 'bicycle', 'car', 'motorcycle', 'airplane', 'bus', 'train', 'truck', 'boat', 'traffic light', 'fire hydrant', 'stop sign', 'parking meter', 'bench', 'bird', 'cat', 'dog', 'horse', 'sheep', 'cow', 'elephant', 'bear', 'zebra', 'giraffe', 'backpack', 'umbrella', 'handbag', 'tie', 'suitcase', 'frisbee', 'skis', 'snowboard', 'sports ball', 'kite', 'baseball bat', 'baseball glove', 'skateboard', 'surfboard', 'tennis racket', 'bottle', 'wine glass', 'cup', 'fork', 'knife', 'spoon', 'bowl', 'banana', 'apple', 'sandwich', 'orange', 'broccoli', 'carrot', 'hot dog', 'pizza', 'donut', 'cake', 'chair', 'couch', 'potted plant', 'bed', 'dining table', 'toilet', 'tv', 'laptop', 'mouse', 'remote', 'keyboard', 'cell phone', 'microwave', 'oven', 'toaster', 'sink', 'refrigerator', 'book', 'clock', 'vase', 'scissors', 'teddy bear', 'hair drier', 'toothbrush']

rng = np.random.default_rng(3)
colors = rng.uniform(0, 255, size=(len(class_names), 3))

# Create a dummy image (height=480, width=640, channels=3)
image = np.ones((480, 640, 3), dtype=np.uint8) * 200

# Create bounding boxes as numpy array
# Format: [x1, y1, x2, y2] for each detection
boxes = np.array([
    [50.0, 100.0, 200.0, 300.0],
    [250.0, 150.0, 450.0, 400.0]
], dtype=np.float32)

# Create scores for each detection
scores = np.array([0.95, 0.87], dtype=np.float32)

# Create class IDs for each detection
class_ids = np.array([0, 2], dtype=np.int32)

# Create mask_alpha parameter
mask_alpha = 0.3

ARGS = {
    'image': image,
    'boxes': boxes,
    'scores': scores,
    'class_ids': class_ids,
    'mask_alpha': mask_alpha
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
cv2, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family int, shape (480, 640, 3) — shape (480, 640, 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
draw_text Ran
ran on a synthesized input
ibaiGorordo/ONNX-YOLOv9-MIT-Object-Detection/yolov9/utils.py
code served (permissive licence) · get_code("8bd590382607dfcb")
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 3, dtype family int, shape (100, 200, 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 cv2
import numpy as np

# Create a simple test image (BGR format, as expected by cv2)
image = np.ones((100, 200, 3), dtype=np.uint8) * 255

# Create a bounding box as numpy array [x1, y1, x2, y2]
box = np.array([10, 20, 80, 60], dtype=np.float32)

# Define text to draw
text = "Test"

# Define color (BGR format for OpenCV)
color = (0, 0, 255)

# Define font size
font_size = 0.5

# Define text thickness
text_thickness = 2

ARGS = {
    "image": image,
    "text": text,
    "box": box,
    "color": color,
    "font_size": font_size,
    "text_thickness": text_thickness
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
cv2, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family int, shape (100, 200, 3) — shape (100, 200, 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
generate_anchors Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
WongKinYiu/YOLO/yolo/utils/bounding_box_utils.py
code served (permissive licence) · get_code("8e61071af5c8f65a")
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 tuple. 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 = {
    "image_size": [320, 240],
    "strides": [8, 16, 32],
}
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 tuple — a tuple output has no array shape of its own; none is 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
lerp Ran
ran, honoured the contract we drafted
WongKinYiu/YOLO/yolo/utils/model_utils.py
code served (permissive licence) · get_code("38e55f346fdfc802")
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 float, 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

ARGS = {
    "start": 0.0,
    "end": 1.0,
    "step": 0.5,
    "total": 1
}
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 float, 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
organize_annotations_by_image Ran
ran; our contract draft was wrong, not the code
WongKinYiu/YOLO/yolo/utils/dataset_utils.py
code served (permissive licence) · get_code("b332cb7d65ce68f1")
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
import numpy as np
from typing import Any, Dict, List, Optional

torch.manual_seed(0)

# Construct test data matching COCO annotation format
data = {
    'annotations': [
        {
            'id': 1,
            'image_id': 101,
            'category_id': 5,
            'iscrowd': False,
            'area': 100.0,
            'bbox': [10, 20, 50, 60]
        },
        {
            'id': 2,
            'image_id': 101,
            'category_id': 3,
            'iscrowd': False,
            'area': 200.0,
            'bbox': [100, 100, 80, 90]
        },
        {
            'id': 3,
            'image_id': 102,
            'category_id': 7,
            'iscrowd': True,
            'area': 500.0,
            'bbox': [0, 0, 100, 100]
        },
        {
            'id': 4,
            'image_id': 102,
            'category_id': 5,
            'iscrowd': False,
            'area': 150.0,
            'bbox': [50, 50, 40, 40]
        }
    ]
}

# Construct id_to_idx mapping for category remapping
id_to_idx = {
    3: 0,
    5: 1,
    7: 2
}

ARGS = {
    'data': data,
    'id_to_idx': id_to_idx
}
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 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
round_up Ran
ran, honoured the contract we drafted
WongKinYiu/YOLO/yolo/utils/module_utils.py
code served (permissive licence) · get_code("8e1ad17d8526e395")
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 int, ndim 0, dtype family int, 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)

# Test with integer input
x_int = 7
div_int = 3

# Test with tensor input
x_tensor = torch.tensor([7, 15, 22], dtype=torch.int32)
div_tensor = 4

ARGS = {
    "x": x_int,
    "div": div_int,
}
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 int, ndim 0, dtype family int, 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
smooth_BCE Ran
ran, honoured the contract we drafted
WongKinYiu/yolov9/utils/loss.py
pointer only (licence: GPL-3.0); the stored copy get_code serves, from Gandolfczjh/APDE, is pointer only (licence: NONE) · get_code("386a04c4b6c96c88")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, WongKinYiu/yolov9, is not in the permissive set Syntology serves inline. 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.

transform_bbox Ran
ran; our contract draft was wrong, not the code
WongKinYiu/YOLO/yolo/utils/bounding_box_utils.py
code served (permissive licence) · get_code("f718ace509af682f")
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 2, dtype family float, shape (2, 4) 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, tensor

torch.manual_seed(0)

# Create a small bounding box tensor
# Shape: (2, 4) - batch of 2 bboxes, 4 coordinates each
# Using xywh format: x, y, width, height
bbox = torch.tensor([
    [10.0, 20.0, 30.0, 40.0],
    [5.0, 15.0, 25.0, 35.0]
], dtype=torch.float32)

ARGS = {
    'bbox': bbox,
    'indicator': 'xywh -> xyxy'
}
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 2, dtype family float, shape (2, 4) — shape (2, 4) 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
RepCSP Not yet run
unverified (harvested, no recorded run)
ultralytics/ultralytics/ultralytics/nn/modules/block.py
pointer only (licence: AGPL-3.0) · get_code("66c4746608dc55b1")
How it ran

The body of this sample is withheld: AGPL-3.0, as GitHub detects it on this paper's repository, ultralytics/ultralytics, is not in the permissive set Syntology serves inline. 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.

RepNCSPELAN Not yet run
unverified (harvested, no recorded run)
henrytsui000/YOLO/yolo/model/module.py
code served (permissive licence) · get_code("2ae37bf73015412d")
How it ran

no invocation recorded.

RepNCSPELAN4 Not yet run
unverified (harvested, no recorded run)
ultralytics/ultralytics/ultralytics/nn/modules/block.py
pointer only (licence: AGPL-3.0) · get_code("c8d57d5b7650e553")
How it ran

The body of this sample is withheld: AGPL-3.0, as GitHub detects it on this paper's repository, ultralytics/ultralytics, is not in the permissive set Syntology serves inline. 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.

RepNCSPELAND Not yet run
unverified (harvested, no recorded run)
henrytsui000/YOLO/yolo/model/module.py
code served (permissive licence) · get_code("5f145be1974d90b9")
How it ran

no invocation recorded.

custom Not yet run
unverified (harvested, no recorded run)
WongKinYiu/yolov9/hubconf.py
pointer only (licence: GPL-3.0); the stored copy get_code serves, from mkang315/PK-YOLO, is pointer only (licence: GPL-3.0) · get_code("6fab24a4d78a99ac")
How it ran

The body of this sample is withheld: GPL-3.0, as GitHub detects it on this paper's repository, WongKinYiu/yolov9, is not in the permissive set Syntology serves inline. 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.

make_ap_table Not yet run
unverified (harvested, no recorded run)
WongKinYiu/YOLO/yolo/utils/solver_utils.py
code served (permissive licence) · get_code("09588c64098f5b10")
How it ran

no invocation recorded.

setup_logger Not yet run
unverified (harvested, no recorded run)
WongKinYiu/YOLO/yolo/utils/logging_utils.py
code served (permissive licence) · get_code("5eb213554929874f")
How it ran

no invocation recorded.

Also present in other papers' code

Of 32 samples lifted from this paper's linked repositories, 32 carry a function identity and 0 do not. Of the 30 distinct functions with an identity, 7 are also present in other papers whose code we hold (4 with at least one paper dated earlier, 1 the same month, 7 later) and 23 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
smooth_BCE 16 other papers
8 dated earlier, 1 the same month, 6 later, 1 same year, order not determinable
1611.02644 (2016-11)
Multispectral Deep Neural Networks for Pedestrian Detection
2108.11250 (2021-08)
YOLOP: You Only Look Once for Panoptic Driving Perception
2202.01811 (2022-02)
ObjectSeeker: Certifiably Robust Object Detection against Patch Hiding Attacks via Patch-agnostic Masking
and 13 more not named here
autopad
brick of this paper's assembly: helper
12 other papers
5 dated earlier, 7 later · 2 only as an assembly brick
1512.03385 (2015-12)
Deep Residual Learning for Image Recognition
1611.02644 (2016-11)
Multispectral Deep Neural Networks for Pedestrian Detection
2004.10934 (2020-04)
YOLOv4: Optimal Speed and Accuracy of Object Detection
and 9 more not named here
custom 5 other papers
3 dated earlier, 2 later
1611.02644 (2016-11)
Multispectral Deep Neural Networks for Pedestrian Detection
2308.11918 (2023-08)
AMSP-UOD: When Vortex Convolution and Stochastic Perturbation Meet Underwater Object Detection
2312.06458 (2023-12)
ASF-YOLO: A Novel YOLO Model with Attentional Scale Sequence Fusion for Cell Instance Segmentation
and 2 more not named here
Conv
brick of this paper's assembly: model
2 other papers
1 dated earlier, 1 later
1512.03385 (2015-12)
Deep Residual Learning for Image Recognition
2504.20670 (2025-04)
FBRT-YOLO: Faster and Better for Real-Time Aerial Image Detection
draw_box 1 other paper
1 later
2405.14458 (2024-05)
YOLOv10: Real-Time End-to-End Object Detection
draw_detections 1 other paper
1 later
2405.14458 (2024-05)
YOLOv10: Real-Time End-to-End Object Detection
draw_text 1 other paper
1 later
2405.14458 (2024-05)
YOLOv10: Real-Time End-to-End Object Detection

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

Today's deep learning methods focus on how to design the objective functions to make the prediction as close as possible to the target. Meanwhile, an appropriate neural network architecture has to be designed. Existing methods ignore a fact that when input data undergoes layer-by-layer feature transformation, large amount of information will be lost. This paper delve into the important issues of information bottleneck and reversible functions. We proposed the concept of programmable gradient information (PGI) to cope with the various changes required by deep networks to achieve multiple objectives. PGI can provide complete input information for the target task to calculate objective function, so that reliable gradient information can be obtained to update network parameters. In addition, a lightweight network architecture -Generalized Efficient Layer Aggregation Network (GELAN) is designed. GELAN confirms that PGI has gained superior results on lightweight models. We verified the proposed GELAN and PGI on MS COCO object detection dataset. The results show that GELAN only uses conventional convolution operators to achieve better parameter utilization than the stateof-the-art methods developed based on depth-wise convolution. PGI can be used for variety of models from lightweight to large. It can be used to obtain complete information, so that train-from-scratch models can achieve better results than state-of-the-art models pre-trained using large datasets, the comparison results are shown in Figure 1. The source codes are released at https://github.com/WongKinYiu/yolov9.

For agents

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

get_harvested_code_for_paper("2402.13616")
get_code_for_paper("2402.13616")
have("2402.13616")

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

curl https://syntology.ai/api/ran/2402.13616.json

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

[![Syntology run record](https://syntology.ai/api/ran/2402.13616.svg)](https://syntology.ai/paper/2402.13616)

Connect an agent — have() is free.