SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2408.06437 · ECCV · 2024

HAT: History-Augmented Anchor Transformer for Online Temporal Action Localization

Octavia Camps, Sakib Reza, Yuexi Zhang, Mohsen Moghaddam

arXiv · PDF · Open in the Atlas

Code that ran

13 of 14 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; 2 honoured the contract we drafted, 0 violated it and 3 have no contract verdict; 8 failures of our instrument; 1 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/2408.06437.json · badge: /api/ran/2408.06437.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
sakibreza/eccv24-hat canonical (classifier label)
classifier basis: census
10 ran · 1 no recorded run
sakibreza/ECCV24-HAT community
Papers with Code archive
3 ran · 0 no recorded run
FunctionStatusWhere it lives
HistoryUnit Ran
ran on a synthesized input · constructed an object; no result was computed
sakibreza/ECCV24-HAT/models.py
code served (permissive licence) · get_code("f03f1bd0ad7a25d8")
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, HistoryUnit: 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 math
import torch.nn as nn

torch.manual_seed(0)

# Define the opt dictionary with all required parameters
ARGS = {
    'opt': {
        'feat_dim': 4,
        'num_of_class': 5,
        'hidden_dim': 8,
        'anchors': [1, 2, 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 HistoryUnit — 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 HistoryUnit 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
MYNET Ran
ran on a synthesized input · constructed an object; no result was computed
sakibreza/ECCV24-HAT/models.py
code served (permissive licence) · get_code("70e832b4db258152")
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, MYNET: 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)

class PositionalEncoding(nn.Module):
    def __init__(self, emb_size: int, dropout: float=0.1, maxlen: int=750):
        super(PositionalEncoding, self).__init__()
        den = torch.exp(-torch.arange(0, emb_size, 2) * math.log(10000) / emb_size)
        pos = torch.arange(0, maxlen).reshape(maxlen, 1)
        pos_embedding = torch.zeros((maxlen, emb_size))
        pos_embedding[:, 0::2] = torch.sin(pos * den)
        pos_embedding[:, 1::2] = torch.cos(pos * den)
        pos_embedding = pos_embedding.unsqueeze(-2)
        self.dropout = nn.Dropout(dropout)
        self.register_buffer('pos_embedding', pos_embedding)

    def forward(self, token_embedding: torch.Tensor):
        return self.dropout(token_embedding + self.pos_embedding[:token_embedding.size(0), :])

class HistoryUnit(torch.nn.Module):
    def __init__(self, opt):
        super(HistoryUnit, self).__init__()
        self.n_feature = opt['feat_dim']
        n_class = opt['num_of_class']
        n_embedding_dim = opt['hidden_dim']
        n_hist_dec_head = 4
        n_hist_dec_layer = 5
        n_hist_dec_head_2 = 4
        n_hist_dec_layer_2 = 2
        self.anchors = opt['anchors']
        self.history_tokens = 16
        self.short_window_size = 16
        self.anchors_stride = []
        dropout = 0.3
        self.best_loss = 1000000
        self.best_map = 0
        self.history_positional_encoding = PositionalEncoding(n_embedding_dim, dropout, maxlen=400)
        self.history_encoder_block1 = nn.TransformerDecoder(nn.TransformerDecoderLayer(d_model=n_embedding_dim, nhead=n_hist_dec_head, dropout=dropout, activation='gelu'), n_hist_dec_layer, nn.LayerNorm(n_embedding_dim))
        self.history_encoder_block2 = nn.TransformerDecoder(nn.TransformerDecoderLayer(d_model=n_embedding_dim, nhead=n_hist_dec_head_2, dropout=dropout, activation='gelu'), n_hist_dec_layer_2, nn.LayerNorm(n_embedding_dim))
        self.snip_head = nn.Sequential(nn.Linear(n_embedding_dim, n_embedding_dim // 4), nn.ReLU())
        self.snip_classifier = nn.Sequential(nn.Linear(self.history_tokens * n_embedding_dim // 4, self.history_tokens * n_embedding_dim // 4 // 4), nn.ReLU(), nn.Linear(self.history_tokens * n_embedding_dim // 4 // 4, n_class))
        self.history_token = nn.Parameter(torch.zeros(self.history_tokens, 1, n_embedding_dim))
        self.norm2 = nn.LayerNorm(n_embedding_dim)
        self.dropout2 = nn.Dropout(0.1)

    def forward(self, long_x, encoded_x):
        hist_pe_x = self.history_positional_encoding(long_x)
        history_token = self.history_token.expand(-1, hist_pe_x.shape[1], -1)
        hist_encoded_x_1 = self.history_encoder_block1(history_token, hist_pe_x)
        hist_encoded_x_2 = self.history_encoder_block2(hist_encoded_x_1, encoded_x)
        hist_encoded_x_2 = hist_encoded_x_2 + self.dropout2(hist_encoded_x_1)
        hist_encoded_x = self.norm2(hist_encoded_x_2)
        snippet_feat = self.snip_head(hist_encoded_x_1)
        snippet_feat = torch.flatten(snippet_feat.permute(1, 0, 2), start_dim=1)
        snip_cls = self.snip_classifier(snippet_feat)
        return (hist_encoded_x, snip_cls)

ARGS = {
    'opt': {
        'feat_dim': 8,
        'num_of_class': 2,
        'hidden_dim': 16,
        'enc_layer': 2,
        'enc_head': 2,
        'dec_layer': 2,
        'dec_head': 2,
        'segment_size': 32,
        'anchors': [0.5, 1.0, 1.5],
    }
}
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 MYNET — 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 MYNET 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
PositionalEncoding Ran
ran on a synthesized input · constructed an object; no result was computed
sakibreza/ECCV24-HAT/models.py
code served (permissive licence) · get_code("14516b65574f5ecf")
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, PositionalEncoding: 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 math
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    'emb_size': 4,
    'dropout': 0.1,
    'maxlen': 750
}
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 PositionalEncoding — 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 PositionalEncoding 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
box_include Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
sakibreza/eccv24-hat/dataset.py
code served (permissive licence) · get_code("3c7815878a239220")
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
import numpy as np

torch.manual_seed(0)

# Analyzing the function:
# - y and target are both accessed with indexing [0] and [1]
# - They appear to be array-like (lists, tuples, or tensors)
# - y[0] - y[1] and y[0] are used in arithmetic operations
# - The function performs comparisons on these values
# - Both y and target should have at least 2 elements

# Create valid arguments:
# y and target should be sequences with at least 2 numeric elements
y = [10, 3]  # y[0] = 10, y[1] = 3, so st = 7, ed = 10
target = [9, 2]  # target[0] = 9, target[1] = 2, so target_st = 7, target_ed = 9

# Verify the logic works:
# st = 10 - 3 = 7
# ed = 10
# target_st = 9 - 2 = 7
# target_ed = 9
# detection_point = 7
# Condition: ed > detection_point (10 > 7 ✓) AND 
#           target_st < st (7 < 7 ✗) 
# So this returns False

# Let's create a case that returns True:
y = [10, 2]  # st = 8, ed = 10
target = [9, 3]  # target_st = 6, target_ed = 9
# Condition: ed > detection_point (10 > 6 ✓) AND 
#           target_st < st (6 < 8 ✓) AND 
#           target_ed > ed (9 > 10 ✗)
# Still False

# Try again:
y = [10, 1]  # st = 9, ed = 10
target = [11, 1]  # target_st = 10, target_ed = 11
# detection_point = 10
# ed > detection_point (10 > 10 ✗)
# Still False

# One more time:
y = [10, 1]  # st = 9, ed = 10
target = [8, 2]  # target_st = 6, target_ed = 8
# detection_point = 6
# ed > detection_point (10 > 6 ✓) AND
# target_st < st (6 < 9 ✓) AND
# target_ed > ed (8 > 10 ✗)
# Still False

# Let's make target_ed > ed:
y = [8, 2]  # st = 6, ed = 8
target = [10, 2]  # target_st = 8, target_ed = 10
# detection_point = 8
# ed > detection_point (8 > 8 ✗)

# Final attempt:
y = [10, 3]  # st = 7, ed = 10
target = [9, 4]  # target_st = 5, target_ed = 9
# detection_point = 5
# ed > detection_point (10 > 5 ✓) AND
# target_st < st (5 < 7 ✓) AND
# target_ed > ed (9 > 10 ✗)

# Make it work:
y = [10, 3]  # st = 7, ed = 10
target = [12, 2]  # target_st = 10, target_ed = 12
# detection_point = 10
# ed > detection_point (10 > 10 ✗)

y = [10, 2]  # st = 8, ed = 10
target = [11, 1]  # target_st = 10, target_ed = 11
# detection_point = 10
# ed > detection_point (10 > 10 ✗)

y = [11, 1]  # st = 10, ed = 11
target = [10, 2]  # target_st = 8, target_ed = 10
# detection_point = 8
# ed > detection_point (11 > 8 ✓) AND
# target_st < st (8 < 10 ✓) AND
# target_ed > ed (10 > 11 ✗)

y = [11, 1]  # st = 10, ed = 11
target = [9, 2]  # target_st = 7, target_ed = 9
# This won't work either

# Simple working case:
y = [10.0, 2.0]
target = [8.0, 1.0]

ARGS = {
    "y": y,
    "target": target
}
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 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
calc_iou Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
sakibreza/eccv24-hat/dataset.py
code served (permissive licence) · get_code("76cad4b96bed8a89")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned float, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# Analyzing the function:
# - calc_iou takes two parameters: a and b
# - Both a and b are indexed with [0] and [1], suggesting they are sequences (lists, tuples, or tensors)
# - a[0] and a[1] are used in arithmetic operations (subtraction, comparison with min/max)
# - b[0] and b[1] are used in arithmetic operations (subtraction, comparison with min/max)
# - The result is a numeric value (iou)

# These appear to be intervals represented as [end, duration] or [position, length]
# Let's use simple numeric values

a = [10, 3]  # [end_position, duration]
b = [12, 4]  # [end_position, duration]

ARGS = {
    "a": a,
    "b": b,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
cls_loss_func Ran
ran; our contract draft was wrong, not the code
sakibreza/eccv24-hat/loss_func.py
code served (permissive licence) · get_code("fa2555ecb9baeda4")
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 0, dtype family float, shape () as a later run recorded it. 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, 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)

# Create small tensors for testing
# y: target tensor with shape (batch_size, num_classes)
# output: model output tensor with shape (batch_size, num_classes)
batch_size = 2
num_classes = 23

y = torch.randint(0, 2, (batch_size, num_classes)).float()
output = torch.randn(batch_size, num_classes)

# weight parameter - can be None or a tensor of weights per class
weight = None

# Build ARGS dict with all parameters for cls_loss_func
ARGS = {
    'y': y,
    'output': output,
    'use_focal': False,
    'weight': weight,
    'reduce': 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 Tensor, 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
cls_loss_func_ Ran
ran; our contract draft was wrong, not the code
sakibreza/eccv24-hat/loss_func.py
code served (permissive licence) · get_code("4010b82fd0123b85")
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 1, dtype family float, shape (2,) as a later run recorded it. 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, 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)

# Create a loss function that takes (output, target) and returns a loss tensor
loss_func = nn.CrossEntropyLoss(reduction='none')

# Create batch tensors
# y: target labels, shape (batch_size=2, num_classes=4)
y = torch.randint(0, 4, (2, 4)).float()

# output: model predictions, shape (batch_size=2, num_classes=4)
output = torch.randn(2, 4)

# weight: class weights, shape (num_classes=4,)
weight = torch.tensor([0.5, 1.0, 1.5, 2.0])

ARGS = {
    'loss_func': loss_func,
    'y': y,
    'output': output,
    'use_focal': False,
    'weight': weight,
    'reduce': 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 Tensor, ndim 1, dtype family float, shape (2,) — shape (2,) 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
get_blocked_videos Ran
ran, honoured the contract we drafted
sakibreza/eccv24-hat/Evaluation/utils.py
code served (permissive licence) · get_code("b976a341da79fc62")
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)

# The function get_blocked_videos only takes an optional 'api' parameter
# The default is already provided as a string constant
# The function body just returns an empty list, so it doesn't actually use the api parameter
# We just need to provide the api parameter with a valid string value

ARGS = {
    'api': 'http://ec2-52-11-11-89.us-west-2.compute.amazonaws.com/challenge17/api.py'
}
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
interpolated_prec_rec Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
sakibreza/eccv24-hat/Evaluation/utils.py
pointer only (licence: MIT); the stored copy get_code serves, from Ontheway361/BSN_pytorch, is pointer only (licence: NONE) · get_code("81aab88321de62c2")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from Ontheway361/BSN_pytorch, 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.

load_json Ran
ran; our contract draft was wrong, not the code
sakibreza/eccv24-hat/dataset.py
code served (permissive licence) · get_code("6feb659b1cc2c1e8")
How it ran

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

import json
import tempfile
import os

# Create a temporary JSON file with valid data
temp_dir = tempfile.mkdtemp()
json_file_path = os.path.join(temp_dir, "test_data.json")

test_data = {
    "key1": "value1",
    "key2": [1, 2, 3],
    "key3": {"nested": "object"}
}

with open(json_file_path, 'w') as f:
    json.dump(test_data, f)

ARGS = {
    "file": json_file_path
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
non_max_suppression Ran
ran; our contract draft was wrong, not the code
sakibreza/eccv24-hat/iou_utils.py
code served (permissive licence) · get_code("44f4e934798b5b3f")
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 numpy as np

# Construct valid arguments for non_max_suppression
proposals = [
    {
        'segment': [0.1, 0.5],
        'score': 0.9,
        'label': 'person'
    },
    {
        'segment': [0.15, 0.55],
        'score': 0.7,
        'label': 'person'
    },
    {
        'segment': [0.6, 0.9],
        'score': 0.85,
        'label': 'car'
    },
    {
        'segment': [0.65, 0.95],
        'score': 0.6,
        'label': 'car'
    },
    {
        'segment': [0.2, 0.3],
        'score': 0.5,
        'label': 'dog'
    }
]

overlapThresh = 0.3

ARGS = {
    'proposals': proposals,
    'overlapThresh': overlapThresh
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — 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
regress_loss_func Ran
ran; our contract draft was wrong, not the code
sakibreza/eccv24-hat/loss_func.py
code served (permissive licence) · get_code("a7c0d044f47c77b5")
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 0, dtype family float, shape () as a later run recorded it. 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, 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)

# The function expects:
# - y: a tensor that will be reshaped and indexed
# - output: a tensor that will be reshaped and indexed
# Both should have at least 2 dimensions (the function reshapes to -1, last_dim)
# y[:, 1] is accessed, so y needs at least 2 columns
# The function compares y[:, 1] < -100.0 to create a mask

# Create small tensors with distinct dimensions
# y shape: (2, 4) - batch of 2, feature dim of 4
y = torch.randn(2, 4)

# output shape: (2, 4) - should match y's shape after reshape
output = torch.randn(2, 4)

# Make sure some values in y[:, 1] are < -100 and some are >= -100
# so both fg and bg splits exist (avoiding edge cases)
y[0, 1] = -150.0  # background
y[1, 1] = 0.0     # foreground

ARGS = {
    'y': y,
    '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 Tensor, 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
segment_iou Ran
ran, honoured the contract we drafted
sakibreza/eccv24-hat/Evaluation/utils.py
pointer only (licence: MIT); the stored copy get_code serves, from Ontheway361/BSN_pytorch, is pointer only (licence: NONE) · get_code("d0744a2fe3151508")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from Ontheway361/BSN_pytorch, 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.

check_overlap_proposal Not yet run
unverified (harvested, no recorded run)
sakibreza/eccv24-hat/iou_utils.py
code served (permissive licence) · get_code("98763476d52320c5")
How it ran

no invocation recorded.

Also present in other papers' code

Of 14 samples lifted from this paper's linked repositories, 14 carry a function identity and 0 do not. Of the 14 distinct functions with an identity, 3 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 0 the same month, 0 later) and 11 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
interpolated_prec_rec 13 other papers
11 dated earlier, 2 same year, order not determinable
1804.05113 (2018-04)
Multilevel Language and Vision Integration for Text-to-Clip Retrieval
1806.02964 (2018-06)
BSN: Boundary Sensitive Network for Temporal Action Proposal Generation
1907.09702 (2019-07)
BMN: Boundary-Matching Network for Temporal Action Proposal Generation
and 10 more not named here
segment_iou 13 other papers
11 dated earlier, 2 same year, order not determinable
1804.05113 (2018-04)
Multilevel Language and Vision Integration for Text-to-Clip Retrieval
1806.02964 (2018-06)
BSN: Boundary Sensitive Network for Temporal Action Proposal Generation
1907.09702 (2019-07)
BMN: Boundary-Matching Network for Temporal Action Proposal Generation
and 10 more not named here
load_json 3 other papers
3 dated earlier
1907.09702 (2019-07)
BMN: Boundary-Matching Network for Temporal Action Proposal Generation
2009.07641 (2020-09)
BSN++: Complementary Boundary Regressor with Scale-Balanced Relation Modeling for Temporal Action Proposal Generation
2204.01680 (2022-04)
TALLFormer: Temporal Action Localization with a Long-memory Transformer

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

Online video understanding often relies on individual frames, leading to frame-by-frame predictions. Recent advancements such as Online Temporal Action Localization (OnTAL), extend this approach to instance-level predictions. However, existing methods mainly focus on short-term context, neglecting historical information. To address this, we introduce the History-Augmented Anchor Transformer (HAT) Framework for OnTAL. By integrating historical context, our framework enhances the synergy between long-term and short-term information, improving the quality of anchor features crucial for classification and localization. We evaluate our model on both procedural egocentric (PREGO) datasets (EGTEA and EPIC) and standard non-PREGO OnTAL datasets (THUMOS and MUSES). Results show that our model outperforms state-of-the-art approaches significantly on PREGO datasets and achieves comparable or slightly superior performance on non-PREGO datasets, underscoring the importance of leveraging long-term history, especially in procedural and egocentric action scenarios. Code is available at: https: //github.com/sakibreza/ECCV24-HAT/.

For agents

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

get_harvested_code_for_paper("2408.06437")
get_code_for_paper("2408.06437")
have("2408.06437")

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

curl https://syntology.ai/api/ran/2408.06437.json

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

[![Syntology run record](https://syntology.ai/api/ran/2408.06437.svg)](https://syntology.ai/paper/2408.06437)

Connect an agent — have() is free.