SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2103.11511 · CVPR · 2021

MoViNets: Mobile Video Networks for Efficient Video Recognition

Boqing Gong, Matthew Brown, Mingxing Tan, Dan Kondratyuk, Liangzhe Yuan, Yandong Li, Li Zhang

arXiv · PDF · Open in the Atlas

Code that ran

9 of 13 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (6 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 0 violated it and 7 have no contract verdict; 1 failure of our instrument; 4 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2103.11511.json · badge: /api/ran/2103.11511.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
Atze00/MoViNet-pytorch community
Papers with Code archive
9 ran · 4 no recorded run
FunctionStatusWhere it lives
Conv2dBNActivation Ran
ran on a synthesized input · constructed an object; no result was computed
Atze00/MoViNet-pytorch/movinets/models.py
code served (permissive licence) · get_code("ac373f664ffe80dd")
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, Conv2dBNActivation: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn
from torch.nn.modules.utils import _pair

torch.manual_seed(0)

ARGS = {
    'in_planes': 3,
    'out_planes': 16,
    'kernel_size': 3,
    'padding': 1,
    'stride': 1,
    'groups': 1,
    'norm_layer': nn.BatchNorm2d,
    'activation_layer': nn.ReLU,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Conv2dBNActivation — 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 Conv2dBNActivation 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
Conv3DBNActivation Ran
ran on a synthesized input · constructed an object; no result was computed
Atze00/MoViNet-pytorch/movinets/models.py
code served (permissive licence) · get_code("2ba2cd56ee729128")
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, Conv3DBNActivation: 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_planes': 3,
    'out_planes': 64,
    'kernel_size': (3, 3, 3),
    'padding': (1, 1, 1),
    'stride': (1, 1, 1),
    'groups': 1,
    'norm_layer': nn.BatchNorm3d,
    'activation_layer': nn.ReLU,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Conv3DBNActivation — 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 Conv3DBNActivation 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
ConvBlock3D Ran
ran on a synthesized input · constructed an object; no result was computed
Atze00/MoViNet-pytorch/movinets/models.py
code served (permissive licence) · get_code("131856849e049bf9")
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, ConvBlock3D: 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 import OrderedDict

torch.manual_seed(0)

# Define stand-in classes for norm_layer and activation_layer
def norm_layer_fn(num_features, eps=0.001):
    return nn.BatchNorm2d(num_features, eps=eps)

def activation_layer_fn():
    return nn.ReLU(inplace=True)

# Build the arguments dictionary
ARGS = {
    'in_planes': 3,
    'out_planes': 16,
    'kernel_size': (3, 3, 3),
    'tf_like': False,
    'causal': False,
    'conv_type': '3d',
    'padding': (1, 1, 1),
    'stride': (1, 1, 1),
    'norm_layer': norm_layer_fn,
    'activation_layer': activation_layer_fn,
    'bias': False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
einops, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ConvBlock3D — 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 ConvBlock3D 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
Hardsigmoid Ran
ran on a synthesized input · constructed an object; no result was computed
Atze00/MoViNet-pytorch/movinets/models.py
code served (permissive licence) · get_code("d81897bae7fd5d1f")
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, Hardsigmoid: 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 Hardsigmoid(nn.Module):
    def __init__(self) -> None:
        super().__init__()

    def forward(self, x: Tensor) -> Tensor:
        x = (0.2 * x + 0.5).clamp(min=0.0, max=1.0)
        return x

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

# Hardsigmoid takes no parameters in __init__, so ARGS should be empty
ARGS = {}

# Test the module
if __name__ == "__main__":
    model = Hardsigmoid(**ARGS)
    output = model(input_tensor)
    print("Input shape:", input_tensor.shape)
    print("Output shape:", output.shape)
    print("Output:", output)
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 Hardsigmoid — 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 Hardsigmoid 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
Swish Ran
ran on a synthesized input · constructed an object; no result was computed
Atze00/MoViNet-pytorch/movinets/models.py
code served (permissive licence) · get_code("c246a91a55da902d")
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, Swish: 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 Swish(nn.Module):

    def __init__(self) -> None:
        super().__init__()

    def forward(self, x: Tensor) -> Tensor:
        return x * torch.sigmoid(x)


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

# Instantiate the module
swish = Swish()

ARGS = {}
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 Swish — 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 Swish 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
TemporalCGAvgPool3D Ran
ran on a synthesized input · constructed an object; no result was computed
Atze00/MoViNet-pytorch/movinets/models.py
code served (permissive licence) · get_code("90c736c9bb473ebf")
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, TemporalCGAvgPool3D: 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)

# TemporalCGAvgPool3D.__init__() takes no parameters beyond self
# The forward method expects input x to be a Tensor with shape (batch, channels, time, height, width)
# Based on the forward method:
# - x.shape is accessed, expecting at least 5 dimensions
# - cumsum is performed on dim=2 (time dimension)
# - divisor is created and expanded to x.shape
# - activation stores cumulative_sum[:, :, -1:] which has shape (batch, channels, 1, height, width)

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

ARGS = {
}
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 TemporalCGAvgPool3D — 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 TemporalCGAvgPool3D 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
_make_divisible Ran
ran, honoured the contract we drafted
Atze00/MoViNet-pytorch/movinets/models.py
code served (permissive licence) · get_code("0cf345647883539d")
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

ARGS = {
    'v': 33.7,
    'divisor': 8,
    'min_value': None
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type int, ndim 0, dtype family int, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
same_padding Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
Atze00/MoViNet-pytorch/movinets/models.py
code served (permissive licence) · get_code("1679ae00114efb03")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 4, dtype family float, shape (2, 1, 5, 5) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
from torch import nn, Tensor

torch.manual_seed(0)

# Create a small input tensor: batch_size=2, channels=1, height=4, width=3
x = torch.randn(2, 1, 4, 3)

in_height = 4
in_width = 3
stride_h = 2
stride_w = 2
filter_height = 3
filter_width = 3

ARGS = {
    "x": x,
    "in_height": in_height,
    "in_width": in_width,
    "stride_h": stride_h,
    "stride_w": stride_w,
    "filter_height": filter_height,
    "filter_width": filter_width,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 4, dtype family float, shape (2, 1, 5, 5) — shape (2, 1, 5, 5) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
tfAvgPool3D Ran
ran on a synthesized input
Atze00/MoViNet-pytorch/movinets/models.py
code served (permissive licence) · get_code("5262209acc985170")
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 5, 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, Tensor
import numpy as np

INIT_ARGS = {}

ARGS = {
    'x': torch.randn(2, 4, 8, 6, 6)  # batch=2, channels=4, depth=8, height=6, width=6
}
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 5, 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)
BasicBneck Not yet run
unverified (harvested, no recorded run)
Atze00/MoViNet-pytorch/movinets/models.py
code served (permissive licence) · get_code("6cee250f62ac6745")
How it ran

no invocation recorded.

CausalModule Not yet run
unverified (harvested, no recorded run)
Atze00/MoViNet-pytorch/movinets/models.py
code served (permissive licence) · get_code("f65eda23dc445262")
How it ran

no invocation recorded.

MoViNet Not yet run
unverified (harvested, no recorded run)
Atze00/MoViNet-pytorch/movinets/models.py
code served (permissive licence) · get_code("c18a67956b8eb0ed")
How it ran

no invocation recorded.

SqueezeExcitation Not yet run
unverified (harvested, no recorded run)
Atze00/MoViNet-pytorch/movinets/models.py
code served (permissive licence) · get_code("633fa35c6d75c2b4")
How it ran

no invocation recorded.

Also present in other papers' code

Of 13 samples lifted from this paper's linked repositories, 13 carry a function identity and 0 do not. Of the 13 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 0 later) and 12 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
_make_divisible
brick of this paper's assembly: helper
1 other paper
1 dated earlier
1905.11946 (2019-05)
EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

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

Abstract

We present Mobile Video Networks (MoViNets), a family of computation and memory efficient video networks that can operate on streaming video for online inference. 3D convolutional neural networks (CNNs) are accurate at video recognition but require large computation and memory budgets and do not support online inference, making them difficult to work on mobile devices. We propose a three-step approach to improve computational efficiency while substantially reducing the peak memory usage of 3D CNNs. First, we design a video network search space and employ neural architecture search to generate efficient and diverse 3D CNN architectures. Second, we introduce the Stream Buffer technique that decouples memory from video clip duration, allowing 3D CNNs to embed arbitrary-length streaming video sequences for both training and inference with a small constant memory footprint. Third, we propose a simple ensembling technique to improve accuracy further without sacrificing efficiency. These three progressive techniques allow MoViNets to achieve state-of-the-art accuracy and efficiency on the Kinetics, Moments in Time, and Charades video action recognition datasets. For instance, MoViNet-A5-Stream achieves the same accuracy as X3D-XL on Kinetics 600 while requiring 80% fewer FLOPs and 65% less memory. Code is available at https: //github.com/google-research/movinet.

For agents

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

get_harvested_code_for_paper("2103.11511")
get_code_for_paper("2103.11511")
have("2103.11511")

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

curl https://syntology.ai/api/ran/2103.11511.json

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

[![Syntology run record](https://syntology.ai/api/ran/2103.11511.svg)](https://syntology.ai/paper/2103.11511)

Connect an agent — have() is free.