SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2201.03545 · CVPR · 2022

A ConvNet for the 2020s

Trevor Darrell, Saining Xie, Chao-Yuan Wu, Zhuang Liu, Christoph Feichtenhofer, Hanzi Mao

arXiv · PDF · Open in the Atlas

Code that ran

57 of 80 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (39 of them constructed an object rather than computing a result; 3 honoured the contract we drafted, 1 violated it and 45 have no contract verdict; 8 failures of our instrument; 23 no recorded run), measured 2026-11-21; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2201.03545.json · badge: /api/ran/2201.03545.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
pytorch/vision community
not in the paper's declared repositories
8 ran · 3 no recorded run
mzeromiko/vmamba community
Papers with Code archive
7 ran · 4 no recorded run
SarthakYadav/audax community
not in the paper's declared repositories
5 ran · 3 no recorded run
frgfm/Holocron community
Papers with Code archive
4 ran · 2 no recorded run
sithu31296/semantic-segmentation community
Papers with Code archive
4 ran · 1 no recorded run
0jason000/convnext community (classifier label)
classifier basis not recorded on the link
4 ran · 1 no recorded run
jmnolte/hccnet community
Papers with Code archive
4 ran · 0 no recorded run
DarshanDeshpande/jax-models community
Papers with Code archive
3 ran · 1 no recorded run
sayakpaul/ConvNeXt-TF community
Papers with Code archive
3 ran · 0 no recorded run
Raghvender1205/ConvNeXt community
Papers with Code archive
3 ran · 0 no recorded run
murufeng/awesome_lightweight_networks community
not in the paper's declared repositories
3 ran · 0 no recorded run
towhee-io/towhee community
not in the paper's declared repositories
2 ran · 1 no recorded run
facebookresearch/ppuda community
Papers with Code archive
2 ran · 1 no recorded run
copy not recorded community
not in the paper's declared repositories
2 ran · 0 no recorded run
bamps53/convnext-tf community
not in the paper's declared repositories
1 ran · 1 no recorded run
mit-han-lab/litepose community
Papers with Code archive
1 ran · 0 no recorded run
waterdisappear/nudt4mstar community
Papers with Code archive
1 ran · 0 no recorded run
k-h-ismail/convnext-dcls community
Papers with Code archive
0 ran · 2 no recorded run
protonx-tf-04-projects/ConvNext-2020s community
Papers with Code archive
0 ran · 2 no recorded run
Owais-Ansari/Unet3plus community
Papers with Code archive
0 ran · 1 no recorded run
FunctionStatusWhere it lives
Block Ran
ran on a synthesized input · constructed an object; no result was computed
towhee-io/towhee/towhee/models/convnext/convnext.py
code served (permissive licence) · get_code("2dd46bec59a83b7b")
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, Block: 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
from torch import nn

torch.manual_seed(0)

ARGS = {
    "dim": 4,
    "drop_path": 0.1,
    "layer_scale_init_value": 1e-6,
}
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 Block — 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 Block 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
Block Ran
ran on a synthesized input · constructed an object; no result was computed
sayakpaul/ConvNeXt-TF/models/convnext.py
code served (permissive licence) · get_code("adc5d3ca9a21de02")
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, Block: 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 timm.models.layers import DropPath

torch.manual_seed(0)

# Define the LayerNorm and Block classes as provided
class LayerNorm(nn.Module):
    def __init__(self, normalized_shape, eps=1e-06, data_format='channels_last'):
        super().__init__()
        self.weight = nn.Parameter(torch.ones(normalized_shape))
        self.bias = nn.Parameter(torch.zeros(normalized_shape))
        self.eps = eps
        self.data_format = data_format
        if self.data_format not in ['channels_last', 'channels_first']:
            raise NotImplementedError
        self.normalized_shape = (normalized_shape,)

    def forward(self, x):
        if self.data_format == 'channels_last':
            import torch.nn.functional as F
            return F.layer_norm(x, self.normalized_shape, self.weight, self.bias, self.eps)
        elif self.data_format == 'channels_first':
            u = x.mean(1, keepdim=True)
            s = (x - u).pow(2).mean(1, keepdim=True)
            x = (x - u) / torch.sqrt(s + self.eps)
            x = self.weight[:, None, None] * x + self.bias[:, None, None]
            return x

class Block(nn.Module):
    def __init__(self, dim, drop_path=0.0, layer_scale_init_value=1e-06):
        super().__init__()
        self.dwconv = nn.Conv2d(dim, dim, kernel_size=7, padding=3, groups=dim)
        self.norm = LayerNorm(dim, eps=1e-06)
        self.pwconv1 = nn.Linear(dim, 4 * dim)
        self.act = nn.GELU()
        self.pwconv2 = nn.Linear(4 * dim, dim)
        self.gamma = nn.Parameter(layer_scale_init_value * torch.ones(dim), requires_grad=True) if layer_scale_init_value > 0 else None
        self.drop_path = DropPath(drop_path) if drop_path > 0.0 else nn.Identity()

    def forward(self, x):
        input = x
        x = self.dwconv(x)
        x = x.permute(0, 2, 3, 1)
        x = self.norm(x)
        x = self.pwconv1(x)
        x = self.act(x)
        x = self.pwconv2(x)
        if self.gamma is not None:
            x = self.gamma * x
        x = x.permute(0, 3, 1, 2)
        x = input + self.drop_path(x)
        return x

# Construct valid arguments for Block
ARGS = {
    'dim': 4,
    'drop_path': 0.1,
    'layer_scale_init_value': 1e-6
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
timm, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Block — 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 Block 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
Block Ran
ran on a synthesized input · constructed an object; no result was computed
Raghvender1205/ConvNeXt/PyTorch/models/convnext.py
pointer only (licence: NONE) · get_code("114ebf6f5300f50a")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, Raghvender1205/ConvNeXt when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

Block Ran
ran on a synthesized input · constructed an object; no result was computed
jmnolte/hccnet/models/convnext3d.py
pointer only (licence: NOASSERTION) · get_code("e5136463e3d9b5d9")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, jmnolte/hccnet (NOASSERTION). 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.

Block Ran
ran on a synthesized input · constructed an object; no result was computed
murufeng/awesome_lightweight_networks/light_cnns/Transformer/ConvNeXt.py
code served (permissive licence) · get_code("0bf3c80a4d223969")
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, Block: 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 = {
    "dim": 4,
    "drop_path": 0.1,
    "layer_scale_init_value": 1e-6,
}
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 Block — 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 Block 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
Block Ran
ran on a synthesized input
bamps53/convnext-tf/models/convnext.py
code served (permissive licence) · get_code("c7a4a3371425c3ed")
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, shape (2, 4, 3, 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 torch
import torch.nn as nn
from timm.models.layers import DropPath

# Import the class
from torch.nn.functional import layer_norm
import torch.nn.functional as F

class LayerNorm(nn.Module):
    """ LayerNorm that supports two data formats: channels_last (default) or channels_first. 
    The ordering of the dimensions in the inputs. channels_last corresponds to inputs with 
    shape (batch_size, height, width, channels) while channels_first corresponds to inputs 
    with shape (batch_size, channels, height, width).
    """

    def __init__(self, normalized_shape, eps=1e-06, data_format='channels_last'):
        super().__init__()
        self.weight = nn.Parameter(torch.ones(normalized_shape))
        self.bias = nn.Parameter(torch.zeros(normalized_shape))
        self.eps = eps
        self.data_format = data_format
        if self.data_format not in ['channels_last', 'channels_first']:
            raise NotImplementedError
        self.normalized_shape = (normalized_shape,)

    def forward(self, x):
        if self.data_format == 'channels_last':
            return F.layer_norm(x, self.normalized_shape, self.weight, self.bias, self.eps)
        elif self.data_format == 'channels_first':
            u = x.mean(1, keepdim=True)
            s = (x - u).pow(2).mean(1, keepdim=True)
            x = (x - u) / torch.sqrt(s + self.eps)
            x = self.weight[:, None, None] * x + self.bias[:, None, None]
            return x

class Block(nn.Module):
    """ ConvNeXt Block. There are two equivalent implementations:
    (1) DwConv -> LayerNorm (channels_first) -> 1x1 Conv -> GELU -> 1x1 Conv; all in (N, C, H, W)
    (2) DwConv -> Permute to (N, H, W, C); LayerNorm (channels_last) -> Linear -> GELU -> Linear; Permute back
    We use (2) as we find it slightly faster in PyTorch
    
    Args:
        dim (int): Number of input channels.
        drop_path (float): Stochastic depth rate. Default: 0.0
        layer_scale_init_value (float): Init value for Layer Scale. Default: 1e-6.
    """

    def __init__(self, dim, drop_path=0.0, layer_scale_init_value=1e-06):
        super().__init__()
        self.dwconv = nn.Conv2d(dim, dim, kernel_size=7, padding=3, groups=dim)
        self.norm = LayerNorm(dim, eps=1e-06)
        self.pwconv1 = nn.Linear(dim, 4 * dim)
        self.act = nn.GELU()
        self.pwconv2 = nn.Linear(4 * dim, dim)
        self.gamma = nn.Parameter(layer_scale_init_value * torch.ones(dim), requires_grad=True) if layer_scale_init_value > 0 else None
        self.drop_path = DropPath(drop_path) if drop_path > 0.0 else nn.Identity()

    def forward(self, x):
        input = x
        x = self.dwconv(x)
        x = x.permute(0, 2, 3, 1)
        x = self.norm(x)
        x = self.pwconv1(x)
        x = self.act(x)
        x = self.pwconv2(x)
        if self.gamma is not None:
            x = self.gamma * x
        x = x.permute(0, 3, 1, 2)
        x = input + self.drop_path(x)
        return x


INIT_ARGS = {
    'dim': 4,
    'drop_path': 0.1,
    'layer_scale_init_value': 1e-6,
}

ARGS = {
    'x': torch.randn(2, 4, 3, 5),
}

if __name__ == '__main__':
    block = Block(**INIT_ARGS)
    output = block(**ARGS)
    print(f"Input shape: {ARGS['x'].shape}")
    print(f"Output shape: {output.shape}")
    print("Block executed successfully!")
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
timm, 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, shape (2, 4, 3, 5) — shape (2, 4, 3, 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
Bottlenext Ran
ran on a synthesized input · constructed an object; no result was computed
frgfm/Holocron/holocron/models/classification/convnext.py
code served (permissive licence) · get_code("c8ac0dd572294b23")
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, Bottlenext: 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
from torch import nn
from functools import partial
from torchvision.ops.stochastic_depth import StochasticDepth

torch.manual_seed(0)

class LayerNorm2d(nn.LayerNorm):
    """Compatibility wrapper of LayerNorm on 2D tensors"""

    def forward(self, x):
        return super().forward(x.permute(0, 2, 3, 1)).permute(0, 3, 1, 2)

class LayerScale(nn.Module):
    """Learnable channel-wise scaling"""

    def __init__(self, chans, scale=1e-06):
        super().__init__()
        self.register_parameter('weight', nn.Parameter(scale * torch.ones(chans)))

    def forward(self, x):
        return x * self.weight.reshape(1, -1, *(1,) * (x.ndim - 2))

class BlurPool2d(nn.Module):
    """Stand-in blur pool layer"""
    def __init__(self, channels, stride=1):
        super().__init__()
        self.stride = stride
        self.channels = channels
    
    def forward(self, x):
        return x

def conv_sequence(in_channels, out_channels, act_layer=None, norm_layer=None, drop_layer=None, conv_layer=None, bn_channels=None, attention_layer=None, blurpool=False, **kwargs):
    """Builds a sequence of convolutional layers."""
    if conv_layer is None:
        conv_layer = nn.Conv2d
    if bn_channels is None:
        bn_channels = out_channels
    conv_stride = kwargs.get('stride', 1)
    if blurpool and conv_stride > 1:
        kwargs['stride'] = 1
    kwargs['bias'] = kwargs.get('bias', norm_layer is None)
    conv_seq = [conv_layer(in_channels, out_channels, **kwargs)]
    if callable(norm_layer):
        conv_seq.append(norm_layer(bn_channels))
    if callable(act_layer):
        conv_seq.append(act_layer)
    if blurpool and conv_stride > 1:
        conv_seq.append(BlurPool2d(bn_channels, stride=conv_stride))
    if callable(attention_layer):
        conv_seq.append(attention_layer(bn_channels))
    if callable(drop_layer):
        conv_seq.append(drop_layer(inplace=True))
    return conv_seq

class _ResBlock(nn.Module):
    expansion = 1

    def __init__(self, convs, downsample=None, act_layer=None):
        super().__init__()
        self.conv = nn.Sequential(*convs)
        self.downsample = downsample
        if isinstance(act_layer, nn.Module):
            self.activation = act_layer

    def forward(self, x):
        identity = x
        out = self.conv(x)
        if self.downsample is not None:
            identity = self.downsample(x)
        out += identity
        if hasattr(self, 'activation'):
            out = self.activation(out)
        return out

class Bottlenext(_ResBlock):
    def __init__(self, inplanes, act_layer=None, norm_layer=None, drop_layer=None, chan_expansion=4, stochastic_depth_prob=0.1, layer_scale=1e-06):
        if norm_layer is None:
            norm_layer = partial(LayerNorm2d, eps=1e-06)
        if act_layer is None:
            act_layer = nn.GELU()
        super().__init__([*conv_sequence(inplanes, inplanes, None, norm_layer, drop_layer, kernel_size=7, padding=3, stride=1, bias=True, groups=inplanes), *conv_sequence(inplanes, inplanes * chan_expansion, act_layer, None, drop_layer, kernel_size=1, stride=1, bias=True), *conv_sequence(inplanes * chan_expansion, inplanes, None, None, drop_layer, kernel_size=1, stride=1, bias=True), LayerScale(inplanes, layer_scale), StochasticDepth(stochastic_depth_prob, 'row')], None, None)

ARGS = {
    'inplanes': 16,
    'act_layer': nn.GELU(),
    'norm_layer': partial(LayerNorm2d, eps=1e-06),
    'drop_layer': None,
    'chan_expansion': 4,
    'stochastic_depth_prob': 0.1,
    'layer_scale': 1e-06,
}
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 Bottlenext — 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 Bottlenext 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
CNBlock Ran
ran on a synthesized input · constructed an object; no result was computed
pytorch/vision/torchvision/models/convnext.py
code served (permissive licence) · get_code("89c908636a8e0602")
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, CNBlock: 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
from torch import nn
from functools import partial

torch.manual_seed(0)

ARGS = {
    "dim": 4,
    "layer_scale": 0.1,
    "stochastic_depth_prob": 0.1,
    "norm_layer": partial(nn.LayerNorm, eps=1e-06),
}
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 CNBlock — 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 CNBlock 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
CNBlockConfig Ran
ran on a synthesized input · constructed an object; no result was computed
pytorch/vision/torchvision/models/convnext.py
code served (permissive licence) · get_code("717211755e50eb05")
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, CNBlockConfig: 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 = {
    'input_channels': 3,
    'out_channels': 64,
    'num_layers': 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 CNBlockConfig — 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 CNBlockConfig 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
Conv2dWrapper Ran
ran on a synthesized input · constructed an object; no result was computed
mzeromiko/vmamba/analyze/convnexts4nd/convnext_timm.py
code served (permissive licence) · get_code("10f409a671e88bc3")
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, Conv2dWrapper: 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 = {
    "dim_in": 3,
    "dim_out": 4,
    "kernel_size": 3,
    "padding": 1,
}
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 Conv2dWrapper — 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 Conv2dWrapper 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
ConvNeXt Ran
ran on a synthesized input · constructed an object; no result was computed
sayakpaul/ConvNeXt-TF/models/convnext.py
code served (permissive licence) · get_code("f4445f68e74841d0")
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, ConvNeXt: 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
import torch.nn.functional as F
from timm.models.layers import trunc_normal_, DropPath

torch.manual_seed(0)

# Define LayerNorm and Block classes as they appear in the function
class LayerNorm(nn.Module):
    """LayerNorm that supports two data formats: channels_last (default) or channels_first."""

    def __init__(self, normalized_shape, eps=1e-06, data_format='channels_last'):
        super().__init__()
        self.weight = nn.Parameter(torch.ones(normalized_shape))
        self.bias = nn.Parameter(torch.zeros(normalized_shape))
        self.eps = eps
        self.data_format = data_format
        if self.data_format not in ['channels_last', 'channels_first']:
            raise NotImplementedError
        self.normalized_shape = (normalized_shape,)

    def forward(self, x):
        if self.data_format == 'channels_last':
            return F.layer_norm(x, self.normalized_shape, self.weight, self.bias, self.eps)
        elif self.data_format == 'channels_first':
            u = x.mean(1, keepdim=True)
            s = (x - u).pow(2).mean(1, keepdim=True)
            x = (x - u) / torch.sqrt(s + self.eps)
            x = self.weight[:, None, None] * x + self.bias[:, None, None]
            return x


class Block(nn.Module):
    """ConvNeXt Block."""

    def __init__(self, dim, drop_path=0.0, layer_scale_init_value=1e-06):
        super().__init__()
        self.dwconv = nn.Conv2d(dim, dim, kernel_size=7, padding=3, groups=dim)
        self.norm = LayerNorm(dim, eps=1e-06)
        self.pwconv1 = nn.Linear(dim, 4 * dim)
        self.act = nn.GELU()
        self.pwconv2 = nn.Linear(4 * dim, dim)
        self.gamma = nn.Parameter(layer_scale_init_value * torch.ones(dim), requires_grad=True) if layer_scale_init_value > 0 else None
        self.drop_path = DropPath(drop_path) if drop_path > 0.0 else nn.Identity()

    def forward(self, x):
        input = x
        x = self.dwconv(x)
        x = x.permute(0, 2, 3, 1)
        x = self.norm(x)
        x = self.pwconv1(x)
        x = self.act(x)
        x = self.pwconv2(x)
        if self.gamma is not None:
            x = self.gamma * x
        x = x.permute(0, 3, 1, 2)
        x = input + self.drop_path(x)
        return x


ARGS = {
    'in_chans': 3,
    'num_classes': 10,
    'depths': [2, 2, 3, 2],
    'dims': [8, 16, 16, 16],
    'drop_path_rate': 0.1,
    'layer_scale_init_value': 1e-6,
    'head_init_scale': 1.0,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
timm, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ConvNeXt — 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 ConvNeXt 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
ConvNeXt Ran
ran on a synthesized input · constructed an object; no result was computed
DarshanDeshpande/jax-models/jax_models/models/convnext.py
code served (permissive licence) · get_code("b2b0744532c3b95e")
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, ConvNeXt: 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 jax
import jax.numpy as jnp
import flax.linen as nn

# Create a small input tensor for testing
# Shape: (batch_size, height, width, channels)
batch_size = 2
height = 32
width = 32
input_channels = 3

inputs = jnp.ones((batch_size, height, width, input_channels), dtype=jnp.float32)

ARGS = {
    'depths': (3, 3, 9, 3),
    'dims': (96, 192, 384, 768),
    'drop_path': 0.0,
    'layer_scale_init_value': 1e-06,
    'head_init_scale': 1.0,
    'attach_head': True,
    'num_classes': 1000,
    'deterministic': None,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
flax, jax — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ConvNeXt — 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 ConvNeXt 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
ConvNeXt Ran
ran on a synthesized input · constructed an object; no result was computed
Raghvender1205/ConvNeXt/PyTorch/models/convnext.py
pointer only (licence: NONE) · get_code("21ca7a5eca0f20a7")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, Raghvender1205/ConvNeXt when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

ConvNeXt Ran
ran on a synthesized input · constructed an object; no result was computed
sithu31296/semantic-segmentation/semseg/models/backbones/convnext.py
code served (permissive licence) · get_code("0a9ccfaec6460f9b")
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, ConvNeXt: 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
from torch import nn, Tensor

class DropPath(nn.Module):
    """Stochastic Depth per sample"""
    def __init__(self, drop_prob=0.0):
        super().__init__()
        self.drop_prob = drop_prob

    def forward(self, x):
        if self.drop_prob == 0.0 or not self.training:
            return x
        keep_prob = 1 - self.drop_prob
        shape = (x.shape[0],) + (1,) * (x.ndim - 1)
        random_tensor = torch.bernoulli(torch.full(shape, keep_prob))
        return x * random_tensor / keep_prob

torch.manual_seed(0)

ARGS = {
    'model_name': 'T'
}
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 ConvNeXt — 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 ConvNeXt 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
ConvNeXt Ran
ran on a synthesized input · constructed an object; no result was computed
murufeng/awesome_lightweight_networks/light_cnns/Transformer/ConvNeXt.py
code served (permissive licence) · get_code("305d76901873eee2")
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, ConvNeXt: 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_chans": 3,
    "num_classes": 10,
    "depths": [2, 2, 2, 2],
    "dims": [64, 128, 256, 512],
    "drop_path_rate": 0.0,
    "layer_scale_init_value": 1e-6,
    "head_init_scale": 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 ConvNeXt — 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 ConvNeXt 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
ConvNeXt3d Ran
ran on a synthesized input · constructed an object; no result was computed
jmnolte/hccnet/models/convnext3d.py
pointer only (licence: NOASSERTION) · get_code("9a81a26d651f9883")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, jmnolte/hccnet (NOASSERTION). 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.

ConvNeXtBlock Ran
ran on a synthesized input · constructed an object; no result was computed
DarshanDeshpande/jax-models/jax_models/models/convnext.py
code served (permissive licence) · get_code("ea89b2384c71ce06")
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, ConvNeXtBlock: 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 jax
import jax.numpy as jnp
import flax.linen as nn
from typing import Optional

# Create a minimal input tensor for the ConvNeXtBlock
# Shape: (batch, height, width, channels) in NHWC format
# Using batch=2, height=8, width=8, channels=4
batch_size = 2
height = 8
width = 8
input_dim = 4

key = jax.random.PRNGKey(0)
inputs = jax.random.normal(key, (batch_size, height, width, input_dim))

ARGS = {
    'dim': 4,
    'layer_scale_init_value': 1e-06,
    'drop_path': 0.1,
    'deterministic': False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
flax, jax — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ConvNeXtBlock — 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 ConvNeXtBlock 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
ConvNeXtBlock Ran
ran on a synthesized input · constructed an object; no result was computed
mzeromiko/vmamba/analyze/convnexts4nd/convnext_timm.py
code served (permissive licence) · get_code("fda3c94b54e0b559")
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, ConvNeXtBlock: 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 timm.models.layers import Mlp, ConvMlp

torch.manual_seed(0)

class DummyNorm(nn.Module):
    def __init__(self, dim):
        super().__init__()
        self.dim = dim
    
    def forward(self, x):
        return nn.functional.layer_norm(x, (self.dim,))

class DummyInstantiate:
    def __call__(self, registry_dict, layer_config, dim):
        # Return a simple Conv2dWrapper-like module
        class DummyLayer(nn.Module):
            def __init__(self, dim):
                super().__init__()
                self.conv = nn.Conv2d(dim, dim, kernel_size=7, padding=3, groups=dim)
            
            def forward(self, x, resolution=None):
                return self.conv(x)
        
        return DummyLayer(dim)

class DummyUtils:
    def __init__(self):
        self.instantiate = DummyInstantiate()

# Build dummy registry and utils
registry = type('Registry', (), {
    'layer': {'test': 'dummy'}
})()

utils = DummyUtils()

# Make utils available globally for the instantiate call
import sys
current_module = sys.modules[__name__]
setattr(current_module, 'utils', utils)
setattr(current_module, 'registry', registry)

ARGS = {
    'dim': 4,
    'drop_path': 0.1,
    'ls_init_value': 1e-6,
    'conv_mlp': False,
    'mlp_ratio': 4,
    'norm_layer': DummyNorm,
    'layer': None,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
timm, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ConvNeXtBlock — 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 ConvNeXtBlock 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
ConvNeXtStage Ran
ran on a synthesized input · constructed an object; no result was computed
SarthakYadav/audax/audax/models/convnext.py
code served (permissive licence) · get_code("9e1c193413b20e6e")
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, ConvNeXtStage: 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 jax
import jax.numpy as jnp
from jax import random
from flax import linen as nn
from functools import partial

# Set seed for reproducibility
key = random.PRNGKey(0)

# Create a small input tensor for ConvNeXtStage
# Shape: (batch_size, height, width, channels)
x = jnp.ones((2, 8, 8, 3), dtype=jnp.float32)

# Define a simple norm_layer callable that returns a LayerNorm module
norm_layer = partial(nn.LayerNorm, dtype=jnp.float32)

# Define dp_rates (drop path rates) for each block in the stage
dp_rates = [0.0, 0.1]

ARGS = {
    'out_channels': 4,
    'stride': 2,
    'depth': 2,
    'dp_rates': dp_rates,
    'ls_init_value': 1.0,
    'norm_layer': norm_layer,
    'cross_stage': False,
    'dtype': jnp.float32,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
flax, jax — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ConvNeXtStage — 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 ConvNeXtStage 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
ConvNormActivation Ran
ran on a synthesized input · constructed an object; no result was computed
pytorch/vision/torchvision/models/convnext.py
code served (permissive licence) · get_code("2748b257d8ff4d3d")
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, ConvNormActivation: 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 collections.abc import Sequence

torch.manual_seed(0)

ARGS = {
    'in_channels': 3,
    'out_channels': 64,
    'kernel_size': 3,
    'stride': 1,
    'padding': None,
    'groups': 1,
    'norm_layer': nn.BatchNorm2d,
    'activation_layer': nn.ReLU,
    'dilation': 1,
    'inplace': True,
    'bias': None,
    'conv_layer': nn.Conv2d,
}
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 ConvNormActivation — 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 ConvNormActivation output the recorded ndim 1 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
DepthwiseConv2D Ran
ran on a synthesized input · constructed an object; no result was computed
DarshanDeshpande/jax-models/jax_models/models/convnext.py
code served (permissive licence) · get_code("b7361cb0c1592f4b")
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, DepthwiseConv2D: 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 jax
import jax.numpy as jnp
import flax.linen as nn
from typing import Sequence, Tuple, Callable, Optional

# Set random seed for reproducibility
jax.config.update('jax_platform_name', 'cpu')

# Create a minimal input tensor for testing
# Shape: (batch=2, height=4, width=3, channels=2)
input_data = jnp.ones((2, 4, 3, 2), dtype=jnp.float32)

# Define the arguments dictionary
ARGS = {
    'kernel_shape': (3, 3),
    'stride': (1, 1),
    'padding': 'SAME',
    'channel_multiplier': 1,
    'use_bias': True,
    'weights_init': nn.initializers.lecun_uniform(),
    'bias_init': nn.initializers.zeros,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
flax, jax — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DepthwiseConv2D — 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 DepthwiseConv2D 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
Downsample Ran
ran on a synthesized input · constructed an object; no result was computed
sithu31296/semantic-segmentation/semseg/models/backbones/convnext.py
code served (permissive licence) · get_code("cfe765a405ec911f")
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, Downsample: 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
from torch import nn

torch.manual_seed(0)

ARGS = {
    "c1": 4,
    "c2": 8,
    "k": 3,
    "s": 1
}
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 Downsample — 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 Downsample output the recorded ndim 1 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
DropPath Ran
ran on a synthesized input · constructed an object; no result was computed
murufeng/awesome_lightweight_networks/light_cnns/Transformer/ConvNeXt.py
code served (permissive licence) · get_code("6255e860ca403ee6")
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, DropPath: 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)

# Create a small input tensor (batch_size=2, channels=4)
x = torch.randn(2, 4)

# Define ARGS dict with all parameters for DropPath.__init__
ARGS = {
    "drop_prob": 0.5,
    "scale_by_keep": True,
}
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 DropPath — 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 DropPath 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
GRN Ran
ran on a synthesized input · constructed an object; no result was computed
jmnolte/hccnet/models/convnext3d.py
pointer only (licence: NOASSERTION) · get_code("28aceeb7bc019d43")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, jmnolte/hccnet (NOASSERTION). 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.

InvBottleneck Ran
ran on a synthesized input
mit-han-lab/litepose/lib/models/layers/layers.py
code served (permissive licence) · get_code("34a4f57109d0223a")
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, shape (2, 4, 8, 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 torch
import torch.nn as nn

def _make_divisible(v, divisor, min_value=None):
    if min_value is None:
        min_value = divisor
    new_v = max(min_value, int(v + divisor / 2) // divisor * divisor)
    if new_v < 0.9 * v:
        new_v += divisor
    return new_v

class InvBottleneck(nn.Module):

    def __init__(self, inplanes, planes, stride=1, ker=3, exp=6):
        super(InvBottleneck, self).__init__()
        feature_dim = _make_divisible(round(inplanes * exp), 8)
        self.inv = nn.Sequential(nn.Conv2d(inplanes, feature_dim, 1, 1, 0, bias=False), nn.BatchNorm2d(feature_dim), nn.ReLU6(inplace=True))
        self.depth_conv = nn.Sequential(nn.Conv2d(feature_dim, feature_dim, ker, stride, ker // 2, groups=feature_dim, bias=False), nn.BatchNorm2d(feature_dim), nn.ReLU6(inplace=True))
        self.point_conv = nn.Sequential(nn.Conv2d(feature_dim, planes, 1, 1, 0, bias=False), nn.BatchNorm2d(planes))
        self.stride = stride
        self.use_residual_connection = stride == 1 and inplanes == planes

    def forward(self, x):
        out = self.inv(x)
        out = self.depth_conv(out)
        out = self.point_conv(out)
        if self.use_residual_connection:
            out += x
        return out

INIT_ARGS = {
    'inplanes': 4,
    'planes': 4,
    'stride': 1,
    'ker': 3,
    'exp': 6
}

ARGS = {
    'x': torch.randn(2, 4, 8, 8)
}

if __name__ == '__main__':
    model = InvBottleneck(**INIT_ARGS)
    model.eval()
    output = model(**ARGS)
    print(f"Output shape: {output.shape}")
    print(f"Output: {output}")
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, shape (2, 4, 8, 8) — shape (2, 4, 8, 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
LayerNorm Ran
ran on a synthesized input · constructed an object; no result was computed
towhee-io/towhee/towhee/models/convnext/convnext.py
code served (permissive licence) · get_code("a0d926a08b9cf439")
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, LayerNorm: 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
from torch import nn

torch.manual_seed(0)

# Test with channels_last format
ARGS = {
    'normalized_shape': 4,
    'eps': 1e-06,
    'data_format': 'channels_last'
}
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 LayerNorm — 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 LayerNorm 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
LayerNorm Ran
ran on a synthesized input · constructed an object; no result was computed
sayakpaul/ConvNeXt-TF/models/convnext.py
code served (permissive licence) · get_code("c34db513de5535f1")
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, LayerNorm: 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)

# The LayerNorm constructor takes:
# - normalized_shape: int or tuple specifying the shape of normalized dimensions
# - eps: float for numerical stability (default 1e-06)
# - data_format: str either 'channels_last' or 'channels_first' (default 'channels_last')

# Looking at the forward method:
# - For channels_last: F.layer_norm is called with normalized_shape as a tuple
# - For channels_first: weight and bias are indexed with [:, None, None], suggesting they're 1D
# The normalized_shape parameter is converted to a tuple in __init__

ARGS = {
    'normalized_shape': 4,
    'eps': 1e-6,
    'data_format': 'channels_last'
}
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 LayerNorm — 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 LayerNorm 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
LayerNorm Ran
ran on a synthesized input · constructed an object; no result was computed
Raghvender1205/ConvNeXt/PyTorch/models/convnext.py
pointer only (licence: NONE) · get_code("61d5fe387e5aea75")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, Raghvender1205/ConvNeXt when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

LayerNorm Ran
ran on a synthesized input · constructed an object; no result was computed
sithu31296/semantic-segmentation/semseg/models/backbones/convnext.py
code served (permissive licence) · get_code("80c7b022ae7942ed")
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, LayerNorm: 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
from torch import nn, Tensor

torch.manual_seed(0)

class LayerNorm(nn.Module):
    """Channel first layer norm
    """

    def __init__(self, normalized_shape, eps=1e-06) -> None:
        super().__init__()
        self.weight = nn.Parameter(torch.ones(normalized_shape))
        self.bias = nn.Parameter(torch.zeros(normalized_shape))
        self.eps = eps

    def forward(self, x: Tensor) -> Tensor:
        u = x.mean(1, keepdim=True)
        s = (x - u).pow(2).mean(1, keepdim=True)
        x = (x - u) / torch.sqrt(s + self.eps)
        x = self.weight[:, None, None] * x + self.bias[:, None, None]
        return x

normalized_shape = 4
eps = 1e-06

ARGS = {
    "normalized_shape": normalized_shape,
    "eps": eps
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type LayerNorm — 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 LayerNorm 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
LayerNorm Ran
ran on a synthesized input · constructed an object; no result was computed
jmnolte/hccnet/models/convnext3d.py
pointer only (licence: NOASSERTION) · get_code("dff029ed85c168ca")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, jmnolte/hccnet (NOASSERTION). 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.

LayerNorm2d Ran
ran on a synthesized input · constructed an object; no result was computed
frgfm/Holocron/holocron/models/classification/convnext.py
code served (permissive licence) · get_code("3851860b73cb71a9")
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, LayerNorm2d: 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)

# LayerNorm2d inherits from nn.LayerNorm
# nn.LayerNorm requires normalized_shape as first positional argument
# LayerNorm2d.forward expects a 4D tensor (batch, channels, height, width)
# It permutes to (batch, height, width, channels), applies LayerNorm, then permutes back
# So normalized_shape should be the number of channels (last dimension after permute)

# Create a small 4D input tensor: (batch=2, channels=4, height=3, width=5)
x = torch.randn(2, 4, 3, 5)

# normalized_shape for LayerNorm should match the last dimension after permutation
# After permute(0, 2, 3, 1): (2, 3, 5, 4), so normalized_shape = 4
normalized_shape = 4

ARGS = {
    "normalized_shape": normalized_shape,
}
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 LayerNorm2d — 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 LayerNorm2d 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
LayerNorm2d Ran
ran on a synthesized input · constructed an object; no result was computed
mzeromiko/vmamba/analyze/convnexts4nd/convnext_timm.py
code served (permissive licence) · get_code("d412faa4209a95a5")
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, LayerNorm2d: 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)

# LayerNorm2d takes normalized_shape and eps as parameters
# normalized_shape should be the channel dimension (C)
# Looking at the forward pass, it permutes (0, 2, 3, 1) so normalized_shape is C
# The weight and bias are indexed with [:, None, None] in the else branch, confirming C is the shape

normalized_shape = 4  # Channel dimension
eps = 1e-6

ARGS = {
    "normalized_shape": normalized_shape,
    "eps": eps,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
timm, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type LayerNorm2d — 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 LayerNorm2d 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
LayerNorm2d Ran
ran on a synthesized input
pytorch/vision/torchvision/models/convnext.py
code served (permissive licence) · get_code("a3e629aa4dc78889")
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 numpy as np

# LayerNorm2d inherits from nn.LayerNorm, so we need to provide
# the arguments that nn.LayerNorm.__init__ expects.
# nn.LayerNorm.__init__ signature:
# def __init__(self, normalized_shape, eps=1e-5, elementwise_affine=True, device=None, dtype=None)

# For LayerNorm2d, normalized_shape should be the channel dimension(s)
# that we want to normalize over after permuting from (B, C, H, W) to (B, H, W, C)

INIT_ARGS = {
    'normalized_shape': (8,),  # Normalizing over 8 channels
    'eps': 1e-5,
    'elementwise_affine': True
}

# The forward method expects a 4D tensor with shape (B, C, H, W)
# where C matches the normalized_shape

ARGS = {
    'x': torch.randn(2, 8, 5, 7)  # batch=2, channels=8, height=5, width=7
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 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)
LayerScale Ran
ran on a synthesized input · constructed an object; no result was computed
frgfm/Holocron/holocron/models/classification/convnext.py
code served (permissive licence) · get_code("c237d46e09539ba8")
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, LayerScale: 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
from torch import nn

torch.manual_seed(0)

ARGS = {
    'chans': 4,
    'scale': 1e-06,
}
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 LayerScale — 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 LayerScale 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
Mlp Ran
ran on a synthesized input · constructed an object; no result was computed
SarthakYadav/audax/audax/models/convnext.py
code served (permissive licence) · get_code("baad12607c91ab0b")
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, Mlp: 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 jax
import jax.numpy as jnp
from jax import random
from flax import linen as nn
from functools import partial
from typing import Callable, Any

def convnext_truncated_init(key, shape, dtype=jnp.float_, std=0.02):
    return random.truncated_normal(key, -2, 2, shape, dtype) * std

default_init = partial(convnext_truncated_init, std=0.02)

ARGS = {
    "hidden_features": 8,
    "out_features": 3,
    "drop": 0.1,
    "activation": nn.gelu,
    "dtype": jnp.float32,
    "kernel_init": default_init,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
flax, jax — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Mlp — 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 Mlp 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
Permute Ran
ran on a synthesized input · constructed an object; no result was computed
pytorch/vision/torchvision/models/convnext.py
code served (permissive licence) · get_code("65b8909854f686af")
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, Permute: 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
from torch import nn

torch.manual_seed(0)

ARGS = {
    "dims": [0, 2, 1, 3]
}
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 Permute — 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 Permute 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
SepConv Ran
ran on a synthesized input
facebookresearch/ppuda/ppuda/deepnets1m/ops.py
code served (permissive licence) · get_code("65848fdb560a8378")
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, shape (2, 3, 8, 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 torch
import torch.nn as nn

# Stand-in implementations for the light versions
class Conv2dLight(nn.Module):
    def __init__(self, in_channels, out_channels, kernel_size, stride=1, padding=0, groups=1, bias=True):
        super(Conv2dLight, self).__init__()
        self.conv = nn.Conv2d(in_channels, out_channels, kernel_size, stride=stride, padding=padding, groups=groups, bias=bias)
    
    def forward(self, x):
        return self.conv(x)

class BatchNorm2dLight(nn.Module):
    def __init__(self, num_features, track_running_stats=True):
        super(BatchNorm2dLight, self).__init__()
        self.bn = nn.BatchNorm2d(num_features, track_running_stats=track_running_stats)
    
    def forward(self, x):
        return self.bn(x)

def bn_layer(norm, C, light):
    if norm in [None, '', 'none']:
        norm_layer = nn.Identity()
    elif norm.startswith('bn'):
        if light:
            norm_layer = BatchNorm2dLight(C, track_running_stats=norm.find('track') >= 0)
        else:
            norm_layer = nn.BatchNorm2d(C, track_running_stats=norm.find('track') >= 0)
    else:
        raise NotImplementedError(norm)
    return norm_layer

def conv_layer(light):
    return Conv2dLight if light else nn.Conv2d

class SepConv(nn.Module):

    def __init__(self, C_in, C_out, ks, stride, padding, norm='bn', light=False):
        super(SepConv, self).__init__()
        self.stride = stride
        self.op = nn.Sequential(nn.ReLU(inplace=False), conv_layer(light)(C_in, C_in, kernel_size=ks, stride=stride, padding=padding, groups=C_in, bias=False), conv_layer(light)(C_in, C_in, kernel_size=1, padding=0, bias=False), bn_layer(norm, C_in, light), nn.ReLU(inplace=False), conv_layer(light)(C_in, C_in, kernel_size=ks, stride=1, padding=padding, groups=C_in, bias=False), conv_layer(light)(C_in, C_out, kernel_size=1, padding=0, bias=False), bn_layer(norm, C_out, light))

    def forward(self, x):
        return self.op(x)

INIT_ARGS = {
    'C_in': 4,
    'C_out': 3,
    'ks': 3,
    'stride': 1,
    'padding': 1,
    'norm': 'bn',
    'light': False
}

ARGS = {
    'x': torch.randn(2, 4, 8, 8)
}

if __name__ == '__main__':
    model = SepConv(**INIT_ARGS)
    output = model(**ARGS)
    print(f"Input shape: {ARGS['x'].shape}")
    print(f"Output shape: {output.shape}")
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, shape (2, 3, 8, 8) — shape (2, 3, 8, 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
Stem Ran
ran on a synthesized input · constructed an object; no result was computed
mzeromiko/vmamba/analyze/convnexts4nd/convnext_timm.py
code served (permissive licence) · get_code("650e4fe6e5133c60")
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, Stem: 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
import math

torch.manual_seed(0)

# Mock utilities and registry since they're not provided
class MockRegistry:
    layer = {}

class MockUtils:
    @staticmethod
    def instantiate(registry_dict, config, *args, **kwargs):
        # Return a simple identity module for mocking
        return nn.Identity()

utils = MockUtils()
registry = MockRegistry()

# Mock norm_layer
def mock_norm_layer(num_features):
    return nn.BatchNorm2d(num_features)

# Mock stem_layer config
mock_stem_layer = {
    'type': 'conv1d',
    'channels': 32
}

# Build ARGS dictionary with all parameters for Stem.__init__
ARGS = {
    'stem_type': 'patch',
    'in_ch': 3,
    'out_ch': 64,
    'img_size': 224,
    'patch_size': 4,
    'stride': 4,
    'stem_channels': 32,
    'stem_layer': mock_stem_layer,
    'stem_l_max': 256,
    'downsample_act': False,
    'downsample_glu': False,
    'norm_layer': mock_norm_layer
}
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 Stem — 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 Stem 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
Stem Ran
ran on a synthesized input · constructed an object; no result was computed
sithu31296/semantic-segmentation/semseg/models/backbones/convnext.py
code served (permissive licence) · get_code("d3287c301a9e5d2a")
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, Stem: 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
from torch import nn

torch.manual_seed(0)

ARGS = {
    "c1": 3,
    "c2": 4,
    "k": 3,
    "s": 1,
}
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 Stem — 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 Stem output the recorded ndim 1 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
StochasticDepth Ran
ran on a synthesized input · constructed an object; no result was computed
pytorch/vision/torchvision/models/convnext.py
code served (permissive licence) · get_code("8e01c115449d22b7")
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, StochasticDepth: 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)

# StochasticDepth.__init__ takes parameters: p and mode
# We need to provide valid values for these parameters

ARGS = {
    "p": 0.5,  # probability between 0 and 1
    "mode": "batch"  # either "batch" or "row"
}
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 StochasticDepth — 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 StochasticDepth 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
TransposedLN Ran
ran on a synthesized input · constructed an object; no result was computed
mzeromiko/vmamba/analyze/convnexts4nd/convnext_timm.py
code served (permissive licence) · get_code("684bcd34dc5acda1")
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, TransposedLN: 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 = {
    'd': 4,
    'scalar': True,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
timm, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type TransposedLN — 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 TransposedLN 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
TransposedLinear Ran
ran on a synthesized input · constructed an object; no result was computed
mzeromiko/vmamba/analyze/convnexts4nd/convnext_timm.py
code served (permissive licence) · get_code("a60bd2bbb558fe05")
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, TransposedLinear: 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
import math

torch.manual_seed(0)

contract = torch.einsum

class TransposedLinear(nn.Module):
    """Linear module on the second-to-last dimension.

    Assumes shape (B, D, L), where L can be 1 or more axis.
    """

    def __init__(self, d_input, d_output, bias=True):
        super().__init__()
        self.weight = nn.Parameter(torch.empty(d_output, d_input))
        nn.init.kaiming_uniform_(self.weight, a=math.sqrt(5))
        if bias:
            self.bias = nn.Parameter(torch.empty(d_output))
            bound = 1 / math.sqrt(d_input)
            nn.init.uniform_(self.bias, -bound, bound)
            setattr(self.bias, '_optim', {'weight_decay': 0.0})
        else:
            self.bias = 0.0

    def forward(self, x):
        num_axis = len(x.shape[2:])
        y = contract('b u ..., v u -> b v ...', x, self.weight) + self.bias.view(-1, *[1] * num_axis)
        return y


ARGS = {
    'd_input': 4,
    'd_output': 3,
    'bias': True,
}
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 TransposedLinear — 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 TransposedLinear 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
_ResBlock Ran
ran on a synthesized input · constructed an object; no result was computed
frgfm/Holocron/holocron/models/classification/convnext.py
code served (permissive licence) · get_code("62ec6c525f2f3b37")
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, _ResBlock: 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 import Tensor

torch.manual_seed(0)

# Build small convolution modules for the convs list
conv1 = nn.Conv2d(4, 8, kernel_size=3, padding=1)
conv2 = nn.Conv2d(8, 4, kernel_size=3, padding=1)
convs = [conv1, conv2]

# Build a downsample module (1x1 conv to match dimensions if needed)
downsample = nn.Sequential(
    nn.Conv2d(4, 4, kernel_size=1),
)

# Build an activation layer
act_layer = nn.ReLU()

ARGS = {
    "convs": convs,
    "downsample": downsample,
    "act_layer": act_layer,
}
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 _ResBlock — 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 _ResBlock 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
_is_contiguous Ran
ran; our contract draft was wrong, not the code
mzeromiko/vmamba/analyze/convnexts4nd/convnext_timm.py
code served (permissive licence) · get_code("b6b433072be7fbfc")
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 torch

torch.manual_seed(0)

# Create a small tensor for testing
tensor = torch.randn(2, 4)

# Ensure the tensor is contiguous
tensor = tensor.contiguous()

ARGS = {
    "tensor": 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 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
_make_ntuple Ran
ran, honoured the contract we drafted
pytorch/vision/torchvision/models/convnext.py
code served (permissive licence) · get_code("793c01e18742fca2")
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) 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 collections
from itertools import repeat

torch_imported = False
try:
    import torch
    torch_imported = True
except ImportError:
    pass

# Test case 1: x is a non-iterable scalar, n is an integer
x_scalar = 5
n_value = 3

# Test case 2: x is an iterable (list), n is an integer
x_iterable = [1, 2, 3, 4]
n_value_2 = 2

# We'll use the first test case for ARGS
ARGS = {
    "x": 5,
    "n": 3
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type 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
bn_layer Ran
ran; our contract draft was wrong, not the code
facebookresearch/ppuda/ppuda/deepnets1m/ops.py
code served (permissive licence) · get_code("5d3e948c34fafbe5")
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 BatchNorm2d. 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)

# Stand-in class for BatchNorm2dLight since it's not in standard PyTorch
class BatchNorm2dLight(nn.Module):
    def __init__(self, num_features, track_running_stats=False):
        super().__init__()
        self.num_features = num_features
        self.track_running_stats = track_running_stats
    
    def forward(self, x):
        return x

# Test cases for different norm values
ARGS = {
    "norm": "bn",
    "C": 16,
    "light": 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 BatchNorm2d — a BatchNorm2d output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a BatchNorm2d 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
constant_init Ran
ran; our contract draft was wrong, not the code
SarthakYadav/audax/audax/models/convnext.py
code served (permissive licence) · get_code("a5e8dd4f5b4997a9")
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 ArrayImpl, ndim 3, dtype family float, shape (2, 4, 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 jax
import jax.numpy as jnp
from jax import dtypes

# Generate a random key for JAX
key = jax.random.PRNGKey(0)

# Define shape as a tuple of integers
shape = (2, 4, 3)

# Define dtype as a JAX dtype
dtype = jnp.float32

# Define constant as a float
constant = 0.04

ARGS = {
    "key": key,
    "shape": shape,
    "dtype": dtype,
    "constant": constant,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
jax — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ArrayImpl, ndim 3, dtype family float, shape (2, 4, 3) — shape (2, 4, 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
convnext_1 Ran
ran on a synthesized input
waterdisappear/nudt4mstar/ATRBench/Classification/ConvNeXt/model/ConvNeXt.py
code served (permissive licence) · get_code("9158c58fe9eb4b02")
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 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 torch
from torch import nn
import numpy as np
from torchvision import models

# Mock torchvision.models.convnext_base to avoid downloads
class MockConvNeXtBase(nn.Module):
    def __init__(self):
        super().__init__()
        self.features = nn.Sequential(
            nn.Conv2d(3, 4, kernel_size=3, stride=1, padding=1),
            nn.BatchNorm2d(4),
            nn.ReLU(),
            nn.Conv2d(4, 8, kernel_size=3, stride=1, padding=1),
            nn.BatchNorm2d(8),
            nn.ReLU(),
        )
        self.avgpool = nn.AdaptiveAvgPool2d((1, 1))
        self.classifier = nn.Sequential(
            nn.LayerNorm([8]),
            nn.Linear(8, 1024),
            nn.Linear(1024, 3),  # Will be replaced
        )

    def forward(self, x):
        x = self.features(x)
        x = self.avgpool(x)
        x = torch.flatten(x, 1)
        x = self.classifier(x)
        return x

# Patch models.convnext_base to return our mock
models.convnext_base = lambda pretrained=False: MockConvNeXtBase()

# Define INIT_ARGS for convnext_1.__init__
INIT_ARGS = {
    'num_classes': 3
}

# Define ARGS for forward method
# Input tensor: batch=2, channels=3, height=4, width=4
ARGS = {
    'x': torch.randn(2, 3, 4, 4)
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
numpy, torch, torchvision — 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 — 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)
convnext_truncated_init Ran
ran; our contract draft was wrong, not the code
SarthakYadav/audax/audax/models/convnext.py
code served (permissive licence) · get_code("af08f1375360cd95")
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 ArrayImpl, ndim 4, dtype family float, shape (4, 3, 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 jax
import jax.numpy as jnp
from jax import random

key = random.PRNGKey(0)
shape = (4, 3, 2, 5)
dtype = jnp.float32
std = 0.02

ARGS = {
    "key": key,
    "shape": shape,
    "dtype": dtype,
    "std": std,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
jax — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ArrayImpl, ndim 4, dtype family float, shape (4, 3, 2, 5) — shape (4, 3, 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
double_conv Ran
ran; our contract draft was wrong, not the code
this paper's copy was not recorded; identical code first harvested from Owais-Ansari/Unet3plus
pointer only · get_code("a505d8b9327be637")
How it ran

The body of this sample is withheld: this paper's copy of the code was not recorded (its link names no repository or source path), so no licence was read for it. 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.

drop_path Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
SarthakYadav/audax/audax/models/convnext.py
code served (permissive licence) · get_code("743bf392200483a3")
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 ArrayImpl, ndim 4, dtype family float, shape (2, 4, 3, 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 jax
import jax.numpy as jnp
from jax import random

# Set seed for reproducibility
key = random.PRNGKey(0)

# Create a small test tensor with shape (batch_size=2, channels=4, height=3, width=3)
# This is a 4D tensor typical for image data
x = jnp.ones((2, 4, 3, 3))

# Create an RNG key for JAX random operations
rng = random.PRNGKey(42)

# Set drop probability
drop_prob = 0.5

ARGS = {
    "x": x,
    "rng": rng,
    "drop_prob": drop_prob,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
jax — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ArrayImpl, ndim 4, dtype family float, shape (2, 4, 3, 3) — shape (2, 4, 3, 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
float_parameter Ran
ran, honoured the contract we drafted
0jason000/convnext/src/transforms.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from chenchenzong/dpc, is pointer only (licence: NONE) · get_code("40d3c39a403ae1f7")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from chenchenzong/dpc, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

init_group_params Ran
ran on a synthesized input
0jason000/convnext/src/optim.py
code served (permissive licence) · get_code("698131638ff8d4ad")
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 torch
import torch.nn as nn

torch.manual_seed(0)

class Param:
    def __init__(self, name):
        self.name = name
        self.data = torch.randn(2, 4)

params = [
    Param('weight'),
    Param('bias'),
    Param('beta'),
    Param('gamma'),
    Param('kernel'),
    Param('scale')
]

weight_decay = 0.01

ARGS = {
    'params': params,
    'weight_decay': weight_decay
}
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 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
int_parameter Ran
ran, honoured the contract we drafted
0jason000/convnext/src/transforms.py
code served (permissive licence) · get_code("e5a370ee6856bd8d")
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
import numpy as np

ARGS = {
    "level": 5,
    "maxval": 100
}
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 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
is_model Ran
ran, violated the contract we drafted
0jason000/convnext/src/model.py
code served (permissive licence) · get_code("b85a6569fd4f3890")
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 torch

# Set up the global model entrypoints dictionary with some test models
_model_entrypoints = {
    'resnet18': 'dummy_model_1',
    'vit_base': 'dummy_model_2',
    'bert_small': 'dummy_model_3'
}

# Define arguments for is_model function
ARGS = {
    'model_name': 'resnet18'
}
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 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
single_conv Ran
ran; our contract draft was wrong, not the code
this paper's copy was not recorded; identical code first harvested from Owais-Ansari/Unet3plus
pointer only · get_code("3649cd96f92a4aa2")
How it ran

The body of this sample is withheld: this paper's copy of the code was not recorded (its link names no repository or source path), so no licence was read for it. 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.

stochastic_depth Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
pytorch/vision/torchvision/models/convnext.py
code served (permissive licence) · get_code("a796360490702186")
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) and returned Tensor, ndim 4, dtype family float, shape (2, 4, 3, 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
from torch import nn, Tensor

torch.manual_seed(0)

# Create a small input tensor with batch size 2 and arbitrary dimensions
input_tensor = torch.randn(2, 4, 3, 3)

# Define parameters for stochastic_depth
p = 0.5  # probability of dropping
mode = "batch"  # can be "batch" or "row"
training = True  # whether to apply stochastic depth

ARGS = {
    "input": input_tensor,
    "p": p,
    "mode": mode,
    "training": training,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 4, dtype family float, shape (2, 4, 3, 3) — shape (2, 4, 3, 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
Block Not yet run
unverified (harvested, no recorded run)
sithu31296/semantic-segmentation/semseg/models/backbones/convnext.py
code served (permissive licence) · get_code("dd9421b72f48ad6d")
How it ran

no invocation recorded.

Block Not yet run
unverified (harvested, no recorded run)
k-h-ismail/convnext-dcls/models/convnext_dcls.py
code served (permissive licence) · get_code("dd371307a76c00b0")
How it ran

no invocation recorded.

Conv2dNormActivation Not yet run
unverified (harvested, no recorded run)
pytorch/vision/torchvision/models/convnext.py
code served (permissive licence) · get_code("9a10ee44c3d37cd1")
How it ran

no invocation recorded.

Showing the first 60. An agent gets every one from get_harvested_code_for_paper("2201.03545").

Also present in other papers' code

Of 80 samples lifted from this paper's linked repositories, 80 carry a function identity and 0 do not. Of the 75 distinct functions with an identity, 23 are also present in other papers whose code we hold (19 with at least one paper dated earlier, 0 the same month, 9 later) and 52 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
is_model 22 other papers
8 dated earlier, 14 later
1409.4842 (2014-09)
Going deeper with convolutions
1512.00567 (2015-12)
Rethinking the Inception Architecture for Computer Vision
1608.06993 (2016-08)
Densely Connected Convolutional Networks
and 19 more not named here
float_parameter 15 other papers
10 dated earlier, 5 later
1409.4842 (2014-09)
Going deeper with convolutions
1512.00567 (2015-12)
Rethinking the Inception Architecture for Computer Vision
1608.06993 (2016-08)
Densely Connected Convolutional Networks
and 12 more not named here
LayerNorm
brick of this paper's assembly: model
14 other papers
1 dated earlier, 13 later · 9 only as an assembly brick
2110.02178 (2021-10)
MobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformer
2301.00808 (2023-01)
ConvNeXt V2: Co-designing and Scaling ConvNets with Masked Autoencoders
2303.15446 (2023-03)
SwiftFormer: Efficient Additive Attention for Transformer-based Real-time Mobile Vision Applications
and 11 more not named here
_log_api_usage_once
brick of this paper's assembly: helper
8 other papers
6 dated earlier, 2 later · 1 only as an assembly brick
1512.03385 (2015-12)
Deep Residual Learning for Image Recognition
1603.09382 (2016-03)
Deep Networks with Stochastic Depth
1905.11946 (2019-05)
EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks
and 5 more not named here
int_parameter 8 other papers
5 dated earlier, 3 later
1409.4842 (2014-09)
Going deeper with convolutions
1512.00567 (2015-12)
Rethinking the Inception Architecture for Computer Vision
1608.06993 (2016-08)
Densely Connected Convolutional Networks
and 5 more not named here
DepthwiseConv2D
brick of this paper's assembly: model
5 other papers
5 dated earlier
2102.12122 (2021-02)
Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutions
2103.14030 (2021-03)
Swin Transformer: Hierarchical Vision Transformer using Shifted Windows
2105.01601 (2021-05)
MLP-Mixer: An all-MLP Architecture for Vision
and 2 more not named here
DropPath
brick of this paper's assembly: helper
5 other papers
5 dated earlier · 1 only as an assembly brick
1603.09382 (2016-03)
Deep Networks with Stochastic Depth
2102.12122 (2021-02)
Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutions
2103.14030 (2021-03)
Swin Transformer: Hierarchical Vision Transformer using Shifted Windows
and 2 more not named here
init_group_params 5 other papers
5 dated earlier
1409.4842 (2014-09)
Going deeper with convolutions
1512.00567 (2015-12)
Rethinking the Inception Architecture for Computer Vision
1608.06993 (2016-08)
Densely Connected Convolutional Networks
and 2 more not named here
model_entrypoint 5 other papers
5 dated earlier
1409.4842 (2014-09)
Going deeper with convolutions
1512.00567 (2015-12)
Rethinking the Inception Architecture for Computer Vision
1608.06993 (2016-08)
Densely Connected Convolutional Networks
and 2 more not named here
stochastic_depth
brick of this paper's assembly: helper
3 other papers
3 dated earlier
1603.09382 (2016-03)
Deep Networks with Stochastic Depth
1905.11946 (2019-05)
EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks
2103.14030 (2021-03)
Swin Transformer: Hierarchical Vision Transformer using Shifted Windows
Block
brick of this paper's assembly: model
2 other papers
2 later · 1 only as an assembly brick
2304.02633 (2023-04)
HNeRV: A Hybrid Neural Representation for Videos
2403.14198 (2024-03)
Unleashing Unlabeled Data: A Paradigm for Cross-View Geo-Localization
Conv2dNormActivation
brick of this paper's assembly: helper
2 other papers
2 dated earlier
1905.11946 (2019-05)
EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks
2010.11929 (2020-10)
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
ConvNormActivation
brick of this paper's assembly: helper
2 other papers
2 dated earlier
1905.11946 (2019-05)
EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks
2010.11929 (2020-10)
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
StochasticDepth
brick of this paper's assembly: helper
2 other papers
2 dated earlier
1905.11946 (2019-05)
EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks
2103.14030 (2021-03)
Swin Transformer: Hierarchical Vision Transformer using Shifted Windows
_make_ntuple
brick of this paper's assembly: helper
2 other papers
2 dated earlier
1905.11946 (2019-05)
EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks
2010.11929 (2020-10)
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
conv_sequence
brick of this paper's assembly: helper
2 other papers
2 dated earlier · 1 only as an assembly brick
1505.04597 (2015-05)
U-Net: Convolutional Networks for Biomedical Image Segmentation
2101.03697 (2021-01)
RepVGG: Making VGG-style ConvNets Great Again
DropPath
brick of this paper's assembly: model
1 other paper
1 later
2403.15139 (2024-03)
Deep Generative Model based Rate-Distortion for Image Downscaling Assessment
Permute
brick of this paper's assembly: helper
1 other paper
1 dated earlier · 1 only as an assembly brick
2103.14030 (2021-03)
Swin Transformer: Hierarchical Vision Transformer using Shifted Windows
TransposedLinear
brick of this paper's assembly: helper
1 other paper
1 later
2608.24597 (2026-08)
Taming foundation model with invariance-oriented pre-training for broad-spectrum EEG analysis across signal-level, brain-state, and brain-health tasks
_is_contiguous
brick of this paper's assembly: helper
1 other paper
1 later
2501.09898 (2025-01)
FoundationStereo: Zero-Shot Stereo Matching

Shown: the 20 functions present in the most other papers. 3 more shared functions are not listed here.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-21.

Repositories linked to this paper

More are linked than shown here.

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

Abstract

The "Roaring 20s" of visual recognition began with the introduction of Vision Transformers (ViTs), which quickly superseded ConvNets as the state-of-the-art image classification model. A vanilla ViT, on the other hand, faces difficulties when applied to general computer vision tasks such as object detection and semantic segmentation. It is the hierarchical Transformers (e.g., Swin Transformers) that reintroduced several ConvNet priors, making Transformers practically viable as a generic vision backbone and demonstrating remarkable performance on a wide variety of vision tasks. However, the effectiveness of such hybrid approaches is still largely credited to the intrinsic superiority of Transformers, rather than the inherent inductive biases of convolutions. In this work, we reexamine the design spaces and test the limits of what a pure ConvNet can achieve. We gradually "modernize" a standard ResNet toward the design of a vision Transformer, and discover several key components that contribute to the performance difference along the way. The outcome of this exploration is a family of pure ConvNet models dubbed Con-vNeXt. Constructed entirely from standard ConvNet modules, ConvNeXts compete favorably with Transformers in terms of accuracy and scalability, achieving 87.8% ImageNet top-1 accuracy and outperforming Swin Transformers on COCO detection and ADE20K segmentation, while maintaining the simplicity and efficiency of standard ConvNets.

For agents

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

get_harvested_code_for_paper("2201.03545")
get_code_for_paper("2201.03545")
have("2201.03545")

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

curl https://syntology.ai/api/ran/2201.03545.json

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

[![Syntology run record](https://syntology.ai/api/ran/2201.03545.svg)](https://syntology.ai/paper/2201.03545)

Connect an agent — have() is free.