SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2211.09992 · NeurIPS · 2022

Look More but Care Less in Video Recognition

Huan Wang, Yun Fu, Yi Xu, Yitian Zhang, Yue Bai

arXiv · PDF · Open in the Atlas

Code that ran

4 of 7 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (3 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 4 have no contract verdict; 0 failures of our instrument; 3 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/2211.09992.json · badge: /api/ran/2211.09992.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
bespontaneous/afnet-pytorch canonical (classifier label)
classifier basis: classifier_2025
4 ran · 3 no recorded run
FunctionStatusWhere it lives
Bottleneck_ample Ran
ran on a synthesized input · constructed an object; no result was computed
bespontaneous/afnet-pytorch/ops/backbone/AF_ResNet.py
code served (permissive licence) · get_code("722eaac491177e21")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, Bottleneck_ample: 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)

# Build a minimal downsample module that the forward method expects
# It should be a Sequential or list-like with potential Conv2d and AvgPool2d
downsample = nn.Sequential(
    nn.Conv2d(64, 256, kernel_size=1, stride=1, bias=False),
    nn.BatchNorm2d(256)
)

# Create input tensor: (batch * num_segments, channels, height, width)
# Using batch=2, num_segments=3, so nt=6
x = torch.randn(6, 64, 8, 8)

# list_little parameter - used as n_segment in TSM
list_little = 3

ARGS = {
    'inplanes': 64,
    'planes': 256,
    'num_segments': 3,
    'stride': 1,
    'downsample': downsample,
    'last_relu': True,
    'patch_groups': 1,
    'base_scale': 2,
    'is_first': 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 Bottleneck_ample — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a Bottleneck_ample 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
GumbleSoftmax Ran
ran on a synthesized input
bespontaneous/afnet-pytorch/ops/backbone/AF_ResNet.py
code served (permissive licence) · get_code("f6de5e19425f73d6")
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 3, dtype family float; no output shape is stated. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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

INIT_ARGS = {
    'hard': False
}

ARGS = {
    'logits': torch.randn(2, 3, 4),
    'temp': 1.0,
    'force_hard': False
}

if __name__ == '__main__':
    from torch.autograd import Variable
    import torch.nn.functional as F
    
    class GumbleSoftmax(torch.nn.Module):

        def __init__(self, hard=False):
            super(GumbleSoftmax, self).__init__()
            self.hard = hard

        def sample_gumbel(self, shape, eps=1e-10):
            """Sample from Gumbel(0, 1)"""
            noise = torch.rand(shape)
            noise.add_(eps).log_().neg_()
            noise.add_(eps).log_().neg_()
            return Variable(noise)

        def sample_gumbel_like(self, template_tensor, eps=1e-10):
            uniform_samples_tensor = template_tensor.clone().uniform_()
            gumble_samples_tensor = -torch.log(eps - torch.log(uniform_samples_tensor + eps))
            return gumble_samples_tensor

        def gumbel_softmax_sample(self, logits, temperature):
            """ Draw a sample from the Gumbel-Softmax distribution"""
            dim = logits.size(-1)
            gumble_samples_tensor = self.sample_gumbel_like(logits.data)
            gumble_trick_log_prob_samples = logits + Variable(gumble_samples_tensor)
            soft_samples = F.softmax(gumble_trick_log_prob_samples / temperature, dim=2)
            return soft_samples

        def gumbel_softmax(self, logits, temperature, hard=False):
            """Sample from the Gumbel-Softmax distribution and optionally discretize.
            Args:
            logits: [batch_size, n_class] unnormalized log-probs
            temperature: non-negative scalar
            hard: if True, take argmax, but differentiate w.r.t. soft sample y
            Returns:
            [batch_size, n_class] sample from the Gumbel-Softmax distribution.
            If hard=True, then the returned sample will be one-hot, otherwise it will
            be a probabilitiy distribution that sums to 1 across classes
            """
            y = self.gumbel_softmax_sample(logits, temperature)
            if hard:
                _, max_value_indexes = y.data.max(2, keepdim=True)
                y_hard = logits.data.clone().zero_().scatter_(2, max_value_indexes, 1)
                y = Variable(y_hard - y.data) + y
            return y

        def forward(self, logits, temp=1, force_hard=False):
            samplesize = logits.size()
            if self.training and (not force_hard):
                return self.gumbel_softmax(logits, temperature=temp, hard=False)
            else:
                return self.gumbel_softmax(logits, temperature=temp, hard=True)

    model = GumbleSoftmax(**INIT_ARGS)
    model.eval()
    output = model(**ARGS)
    print("Output shape:", output.shape)
    print("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 3, 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)
dynamic_fusion Ran
ran on a synthesized input · constructed an object; no result was computed
bespontaneous/afnet-pytorch/ops/backbone/AF_ResNet.py
code served (permissive licence) · get_code("05db6c4ace48af9a")
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, dynamic_fusion: 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 dynamic_fusion module takes a channel parameter and optional reduction parameter
# Looking at the forward methods:
# - x is expected to be a 4D tensor (b, c, h, w)
# - The module applies adaptive average pooling, then passes through FC layers
# - Output is attention weights of shape (b, c, 1, 1)

channel = 16
reduction = 16

# Create a small input tensor for testing
# batch_size=2, channels=16, height=4, width=3
x = torch.randn(2, 16, 4, 3)

ARGS = {
    'channel': channel,
    'reduction': reduction
}
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 dynamic_fusion — 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 dynamic_fusion 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
navigation Ran
ran on a synthesized input · constructed an object; no result was computed
bespontaneous/afnet-pytorch/ops/backbone/AF_ResNet.py
code served (permissive licence) · get_code("5a5b1b97c63d9eb6")
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, navigation: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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 navigation module expects:
# - inplanes: number of input channels (default 64)
# - num_segments: number of segments (default 8)
# We need to construct a navigation instance with these parameters

ARGS = {
    "inplanes": 64,
    "num_segments": 8,
}
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 navigation — 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 navigation 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
AFModule Not yet run
unverified (harvested, no recorded run)
bespontaneous/afnet-pytorch/ops/backbone/AF_ResNet.py
code served (permissive licence) · get_code("5d8c29317a714dd1")
How it ran

no invocation recorded.

AFResNet Not yet run
unverified (harvested, no recorded run)
bespontaneous/afnet-pytorch/ops/backbone/AF_ResNet.py
code served (permissive licence) · get_code("1b8764ca6239ee81")
How it ran

no invocation recorded.

TSM Not yet run
unverified (harvested, no recorded run)
bespontaneous/afnet-pytorch/ops/backbone/AF_ResNet.py
code served (permissive licence) · get_code("6789d3209771b4e6")
How it ran

no invocation recorded.

Also present in other papers' code

Of 7 samples lifted from this paper's linked repositories, 7 carry a function identity and 0 do not. Of the 7 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 7 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.

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

Existing action recognition methods typically sample a few frames to represent each video to avoid the enormous computation, which often limits the recognition performance. To tackle this problem, we propose Ample and Focal Network (AFNet), which is composed of two branches to utilize more frames but with less computation. Specifically, the Ample Branch takes all input frames to obtain abundant information with condensed computation and provides the guidance for Focal Branch by the proposed Navigation Module; the Focal Branch squeezes the temporal size to only focus on the salient frames at each convolution block; in the end, the results of two branches are adaptively fused to prevent the loss of information. With this design, we can introduce more frames to the network but cost less computation. Besides, we demonstrate AFNet can utilize fewer frames while achieving higher accuracy as the dynamic selection in intermediate features enforces implicit temporal modeling. Further, we show that our method can be extended to reduce spatial redundancy with even less cost. Extensive experiments on five datasets demonstrate the effectiveness and efficiency of our method. Our code is available at https://github.com/BeSpontaneous/AFNet-pytorch.

For agents

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

get_harvested_code_for_paper("2211.09992")
get_code_for_paper("2211.09992")
have("2211.09992")

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

curl https://syntology.ai/api/ran/2211.09992.json

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

[![Syntology run record](https://syntology.ai/api/ran/2211.09992.svg)](https://syntology.ai/paper/2211.09992)

Connect an agent — have() is free.