SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2204.01692 · ECCV · 2022

Long Movie Clip Classification with State-Space Video Models

Gedas Bertasius, Mohaiminul Islam

arXiv · PDF · Open in the Atlas

Code that ran

11 of 18 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (2 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 0 violated it and 2 have no contract verdict; 8 failures of our instrument; 7 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2204.01692.json · badge: /api/ran/2204.01692.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
md-mohaiminul/ViS4mer marked official in Papers with Code
Papers with Code archive
9 ran · 7 no recorded run
copy not recorded community
not in the paper's declared repositories
2 ran · 0 no recorded run
FunctionStatusWhere it lives
Activation Ran
ran; our contract draft was wrong, not the code
this paper's copy was not recorded; identical code first harvested from raminmh/liquid-s4
pointer only · get_code("89b10a8b26bbb7ed")
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.

LinearActivation Ran
ran; our contract draft was wrong, not the code
md-mohaiminul/ViS4mer/models.py
code served (permissive licence) · get_code("8538f8c63d35c959")
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 Sequential. 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 functools import partial
import math
import opt_einsum as oe

torch.manual_seed(0)

def Activation(activation=None, dim=-1):
    if activation in [None, 'id', 'identity', 'linear']:
        return nn.Identity()
    elif activation == 'tanh':
        return nn.Tanh()
    elif activation == 'relu':
        return nn.ReLU()
    elif activation == 'gelu':
        return nn.GELU()
    elif activation in ['swish', 'silu']:
        return nn.SiLU()
    elif activation == 'glu':
        return nn.GLU(dim=dim)
    elif activation == 'sigmoid':
        return nn.Sigmoid()
    else:
        raise NotImplementedError("hidden activation '{}' is not implemented".format(activation))

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

    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, 1))
            bound = 1 / math.sqrt(d_input)
            nn.init.uniform_(self.bias, -bound, bound)
        else:
            self.bias = 0.0

    def forward(self, x):
        contract = oe.contract
        return contract('... u l, v u -> ... v l', x, self.weight) + self.bias

contract = oe.contract

def get_initializer(name, activation=None):
    if activation in [None, 'id', 'identity', 'linear', 'modrelu']:
        nonlinearity = 'linear'
    elif activation in ['relu', 'tanh', 'sigmoid']:
        nonlinearity = activation
    elif activation in ['gelu', 'swish', 'silu']:
        nonlinearity = 'relu'
    else:
        raise NotImplementedError(f'get_initializer: activation {activation} not supported')
    if name == 'uniform':
        initializer = partial(torch.nn.init.kaiming_uniform_, nonlinearity=nonlinearity)
    elif name == 'normal':
        initializer = partial(torch.nn.init.kaiming_normal_, nonlinearity=nonlinearity)
    elif name == 'xavier':
        initializer = torch.nn.init.xavier_normal_
    elif name == 'zero':
        initializer = partial(torch.nn.init.constant_, val=0)
    elif name == 'one':
        initializer = partial(torch.nn.init.constant_, val=1)
    else:
        raise NotImplementedError(f'get_initializer: initializer type {name} not supported')
    return initializer

def LinearActivation(d_input, d_output, bias=True, zero_bias_init=False, transposed=False, initializer=None, activation=None, activate=False, weight_norm=False, **kwargs):
    """ Returns a linear nn.Module with control over axes order, initialization, and activation """
    linear_cls = TransposedLinear if transposed else nn.Linear
    if activation == 'glu':
        d_output *= 2
    linear = linear_cls(d_input, d_output, bias=bias, **kwargs)
    if initializer is not None:
        get_initializer(initializer, activation)(linear.weight)
    if bias and zero_bias_init:
        nn.init.zeros_(linear.bias)
    if weight_norm:
        linear = nn.utils.weight_norm(linear)
    if activate and activation is not None:
        activation = Activation(activation, dim=-2 if transposed else -1)
        linear = nn.Sequential(linear, activation)
    return linear

ARGS = {
    'd_input': 4,
    'd_output': 3,
    'bias': True,
    'zero_bias_init': False,
    'transposed': False,
    'initializer': 'uniform',
    'activation': 'relu',
    'activate': True,
    'weight_norm': False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
opt_einsum, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential 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
SSKernelSlow Ran
ran on a synthesized input · constructed an object; no result was computed
md-mohaiminul/ViS4mer/models.py
code served (permissive licence) · get_code("39159b1078607edf")
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, SSKernelSlow: 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 omegaconf import DictConfig

torch.manual_seed(0)

# Set up parameters for SSKernelSlow
L = 8  # sequence length
N = 4  # state dimension

# Create A matrix (N x N)
A = torch.randn(1, N, N, dtype=torch.complex64)

# Create B vector (N,)
B = torch.randn(1, N, dtype=torch.complex64)

# Create C matrix with batch dimension (batch=2, N)
C = torch.randn(2, N, dtype=torch.complex64)

# Create log_dt scalar (1,)
log_dt = torch.tensor([0.1], dtype=torch.float32)

# Create trainable config
trainable = DictConfig({
    'A': 0,
    'B': 0,
    'C': 0,
    'dt': 0
})

# Create lr config
lr = DictConfig({
    'A': None,
    'B': None,
    'C': None,
    'dt': None
})

ARGS = {
    'L': L,
    'A': A,
    'B': B,
    'C': C,
    'log_dt': log_dt,
    'trainable': trainable,
    'lr': lr
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
einops, omegaconf, opt_einsum, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type SSKernelSlow — 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 SSKernelSlow 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
md-mohaiminul/ViS4mer/models.py
code served (permissive licence) · get_code("71048363ae476934")
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
import opt_einsum as oe

torch.manual_seed(0)

# Test setup for TransposedLinear
d_input = 4
d_output = 3
bias = True

ARGS = {
    'd_input': d_input,
    'd_output': d_output,
    'bias': bias
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
opt_einsum, 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
get_initializer Ran
ran; our contract draft was wrong, not the code
this paper's copy was not recorded; identical code first harvested from ctlllll/sgconv
pointer only; the stored copy get_code serves, from ctlllll/sgconv, is pointer only (licence: NONE) · get_code("49d9214ec79a8edb")
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.

get_logger Ran
ran; our contract draft was wrong, not the code
md-mohaiminul/ViS4mer/models.py
code served (permissive licence) · get_code("61139ec62260b411")
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 Logger. 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 logging
from pytorch_lightning.utilities import rank_zero_only

# Construct arguments for get_logger function
ARGS = {
    "name": "test_logger",
    "level": logging.DEBUG
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pytorch_lightning — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Logger — a Logger output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Logger 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
krylov Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
md-mohaiminul/ViS4mer/models.py
code served (permissive licence) · get_code("030ade7f4f951898")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 2, dtype family float, shape (2, 3) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

# Set up small tensors for the krylov function
batch_size = 2
matrix_dim = 4
L = 3

# b: input vector, shape (..., n)
b = torch.randn(batch_size, matrix_dim)

# A: square matrix, shape (..., n, n)
A = torch.randn(batch_size, matrix_dim, matrix_dim)

# c: optional weighting vector, shape (..., n)
c = torch.randn(batch_size, matrix_dim)

# Test both with and without return_power
# Here we'll use return_power=False for this setup
return_power = False

ARGS = {
    "L": L,
    "A": A,
    "b": b,
    "c": c,
    "return_power": return_power
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (2, 3) — shape (2, 3) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
nplr Ran
ran; our contract draft was wrong, not the code
md-mohaiminul/ViS4mer/models.py
code served (permissive licence) · get_code("52f5a73f3505ecee")
How it ran

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

import torch
import numpy as np
from scipy import special as ss
import opt_einsum as oe

torch.manual_seed(0)

ARGS = {
    "measure": "legs",
    "N": 4,
    "rank": 1,
    "dtype": torch.float,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, opt_einsum, scipy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
power Ran
ran; our contract draft was wrong, not the code
md-mohaiminul/ViS4mer/models.py
code served (permissive licence) · get_code("fae8dcc7964cac77")
How it ran

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

import torch
from einops import rearrange, repeat

torch.manual_seed(0)

# Construct arguments for the power function
# power(L, A, v=None)
# L: int - the exponent
# A: (..., N, N) - a square matrix
# v: (..., N, L) optional - a matrix for the scan computation

# Create a small square matrix A with shape (N, N)
N = 4
A = torch.randn(N, N)

# L should be a positive integer for the exponent computation
L = 8

# Create v with shape (..., N, L)
# v should have last two dimensions (N, L) where N matches A's dimension
# and L matches the parameter L
v = torch.randn(N, L)

ARGS = {
    "L": L,
    "A": A,
    "v": v,
}
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 tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
rank_correction Ran
ran, honoured the contract we drafted
md-mohaiminul/ViS4mer/models.py
code served (permissive licence) · get_code("3dce80b93036d192")
How it ran

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

import torch

torch.manual_seed(0)

ARGS = {
    "measure": "legs",
    "N": 4,
    "rank": 1,
    "dtype": torch.float
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (1, 4) — shape (1, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
transition Ran
ran; our contract draft was wrong, not the code
md-mohaiminul/ViS4mer/models.py
code served (permissive licence) · get_code("87ea80b70e1e70b9")
How it ran

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

import numpy as np
from scipy import special as ss

# Test with different measures
ARGS = {
    'measure': 'legt',
    'N': 4,
    'beta': 1.0,
    'alpha': 0.0,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
HippoSSKernel Not yet run
unverified (harvested, no recorded run)
md-mohaiminul/ViS4mer/models.py
code served (permissive licence) · get_code("717d9447a001c75e")
How it ran

no invocation recorded.

OptimModule Not yet run
unverified (harvested, no recorded run)
md-mohaiminul/ViS4mer/models.py
code served (permissive licence) · get_code("8076f48bc17f0bc5")
How it ran

no invocation recorded.

S4 Not yet run
unverified (harvested, no recorded run)
md-mohaiminul/ViS4mer/models.py
code served (permissive licence) · get_code("0bbcf5a8797deb01")
How it ran

no invocation recorded.

SSKernelNPLR Not yet run
unverified (harvested, no recorded run)
md-mohaiminul/ViS4mer/models.py
code served (permissive licence) · get_code("a0cd8021dee737f7")
How it ran

no invocation recorded.

ViS4mer Not yet run
unverified (harvested, no recorded run)
md-mohaiminul/ViS4mer/models.py
code served (permissive licence) · get_code("b7f3ea1de74c55c3")
How it ran

no invocation recorded.

_broadcast_dims Not yet run
unverified (harvested, no recorded run)
md-mohaiminul/ViS4mer/models.py
code served (permissive licence) · get_code("0c3303405c8151e9")
How it ran

no invocation recorded.

cauchy_conj Not yet run
unverified (harvested, no recorded run)
md-mohaiminul/ViS4mer/models.py
code served (permissive licence) · get_code("fba66bbdebdb89e4")
How it ran

no invocation recorded.

Also present in other papers' code

Of 18 samples lifted from this paper's linked repositories, 18 carry a function identity and 0 do not. Of the 18 distinct functions with an identity, 8 are also present in other papers whose code we hold (7 with at least one paper dated earlier, 0 the same month, 6 later) and 10 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
Activation
brick of this paper's assembly: helper
15 other papers
6 dated earlier, 9 later · 2 only as an assembly brick
1312.5602 (2013-12)
Playing Atari with Deep Reinforcement Learning
2008.07669 (2020-08)
HiPPO: Recurrent Memory with Optimal Polynomial Projections
2009.09761 (2020-09)
DiffWave: A Versatile Diffusion Model for Audio Synthesis
and 12 more not named here
get_logger
brick of this paper's assembly: helper
15 other papers
2 dated earlier, 13 later · 3 only as an assembly brick
2009.09761 (2020-09)
DiffWave: A Versatile Diffusion Model for Audio Synthesis
2111.00396 (2021-11)
Efficiently Modeling Long Sequences with Structured State Spaces
2205.08897 (2022-05)
FiLM: Frequency improved Legendre Memory Model for Long-term Time Series Forecasting
and 12 more not named here
get_initializer
brick of this paper's assembly: helper
10 other papers
4 dated earlier, 6 later · 1 only as an assembly brick
2008.07669 (2020-08)
HiPPO: Recurrent Memory with Optimal Polynomial Projections
2111.00396 (2021-11)
Efficiently Modeling Long Sequences with Structured State Spaces
2202.09729 (2022-02)
It's Raw! Audio Generation with State-Space Models
and 7 more not named here
power
brick of this paper's assembly: helper
4 other papers
3 dated earlier, 1 later
1312.5602 (2013-12)
Playing Atari with Deep Reinforcement Learning
2009.09761 (2020-09)
DiffWave: A Versatile Diffusion Model for Audio Synthesis
2111.00396 (2021-11)
Efficiently Modeling Long Sequences with Structured State Spaces
and 1 more not named here
krylov
brick of this paper's assembly: helper
2 other papers
1 dated earlier, 1 later
2110.13985 (2021-10)
Combining Recurrent, Convolutional, and Continuous-time Models with Linear State-Space Layers
2411.04165 (2024-11)
Bio-xLSTM: Generative modeling, representation and in-context learning of biological and chemical sequences
LinearActivation
brick of this paper's assembly: helper
1 other paper
1 dated earlier
2111.00396 (2021-11)
Efficiently Modeling Long Sequences with Structured State Spaces
TransposedLinear
brick of this paper's assembly: helper
1 other paper
1 dated earlier
2111.00396 (2021-11)
Efficiently Modeling Long Sequences with Structured State Spaces
_broadcast_dims
brick of this paper's assembly: helper
1 other paper
1 later · 1 only as an assembly brick
2212.14052 (2022-12)
Hungry Hungry Hippos: Towards Language Modeling with State Space Models

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

Most modern video recognition models are designed to operate on short video clips (e.g., 5-10s in length). Thus, it is challenging to apply such models to long movie understanding tasks, which typically require sophisticated long-range temporal reasoning. The recently introduced video transformers partially address this issue by using longrange temporal self-attention. However, due to the quadratic cost of selfattention, such models are often costly and impractical to use. Instead, we propose ViS4mer, an efficient long-range video model that combines the strengths of self-attention and the recently introduced structured state-space sequence (S4) layer. Our model uses a standard Transformer encoder for short-range spatiotemporal feature extraction, and a multiscale temporal S4 decoder for subsequent long-range temporal reasoning. By progressively reducing the spatiotemporal feature resolution and channel dimension at each decoder layer, ViS4mer learns complex longrange spatiotemporal dependencies in a video. Furthermore, ViS4mer is 2.63× faster and requires 8× less GPU memory than the corresponding pure self-attention-based model. Additionally, ViS4mer achieves state-ofthe-art results in 6 out of 9 long-form movie video classification tasks on the Long Video Understanding (LVU) benchmark. Furthermore, we show that our approach successfully generalizes to other domains, achieving competitive results on the Breakfast and the COIN procedural activity datasets. The code is publicly available. 1

For agents

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

get_harvested_code_for_paper("2204.01692")
get_code_for_paper("2204.01692")
have("2204.01692")

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

curl https://syntology.ai/api/ran/2204.01692.json

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

[![Syntology run record](https://syntology.ai/api/ran/2204.01692.svg)](https://syntology.ai/paper/2204.01692)

Connect an agent — have() is free.