SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2104.00990 · CVPR · 2021

Visual Semantic Role Labeling for Video Understanding

Aniruddha Kembhavi, Arka Sadhu, Mark Yatskar, Ram Nevatia, Tanmay Gupta

arXiv · PDF · Open in the Atlas

Code that ran

9 of 9 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 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; 5 failures of our instrument; 0 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/2104.00990.json · badge: /api/ran/2104.00990.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
TheShadow29/VidSitu canonical (classifier label)
classifier basis: census
9 ran · 0 no recorded run
FunctionStatusWhere it lives
arg_mapper Ran
ran; our contract draft was wrong, not the code
TheShadow29/VidSitu/vidsitu_code/evl_fns.py
code served (permissive licence) · get_code("bb2ee2eda144860b")
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 str. 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 re

def arg_mapper(arg_inp, argm_re=None):
    if argm_re is None:
        argm_re = re.compile('ArgM (.*)')
    arg_name = arg_inp.split(' ')[0]
    if arg_name in set(['Arg0', 'Arg1', 'Arg2', 'Arg3', 'Arg4', 'Arg5']):
        return arg_name
    elif arg_inp == 'Scene of the Event':
        return 'AScn'
    else:
        assert arg_name == 'ArgM'
        y2 = argm_re.findall(arg_inp)[0].strip()
        if 'direction' in y2:
            return 'ADir'
        elif 'purpose' in y2:
            return 'APrp'
        elif 'manner' in y2:
            return 'AMnr'
        elif 'location' in y2:
            return 'ALoc'
        elif 'goal' in y2:
            return 'AGol'
        else:
            raise NotImplementedError

ARGS = {
    'arg_inp': 'Arg0 some argument',
    'argm_re': re.compile('ArgM (.*)')
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
collate_dct_lst_naive Ran
ran; our contract draft was wrong, not the code
TheShadow29/VidSitu/vidsitu_code/evl_fns.py
code served (permissive licence) · get_code("783738a17e5778da")
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 torch
import numpy as np

torch.manual_seed(0)

# The function collate_dct_lst_naive takes a list of dictionaries
# and reorganizes them so each key maps to a list of values
# It expects batch to be a List[Dict]

batch = [
    {
        "input_ids": torch.tensor([1, 2, 3, 4]),
        "attention_mask": torch.tensor([1, 1, 1, 0]),
        "label": 0,
    },
    {
        "input_ids": torch.tensor([5, 6, 7, 8]),
        "attention_mask": torch.tensor([1, 1, 1, 1]),
        "label": 1,
    },
]

ARGS = {
    "batch": batch,
}
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 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
enclose_ag Ran
ran on a synthesized input
TheShadow29/VidSitu/vidsitu_code/dat_loader.py
code served (permissive licence) · get_code("0fd9228e4e9231e0")
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 str. 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

def st_ag(ag):
    return f'<{ag}>'

def end_ag(ag):
    return f'</{ag}>'

def enclose_ag(agname, ag_str):
    return f'{st_ag(agname)} {ag_str} {end_ag(agname)}'

ARGS = {
    'agname': 'div',
    'ag_str': 'Hello World'
}
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
end_ag Ran
ran on a synthesized input
TheShadow29/VidSitu/vidsitu_code/dat_loader.py
code served (permissive licence) · get_code("b32761c4fc18d6c4")
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 str. 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

def end_ag(ag):
    return f'</{ag}>'

ARGS = {
    'ag': 'div'
}
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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_name_from_inst Ran
ran; our contract draft was wrong, not the code
TheShadow29/VidSitu/main_dist.py
code served (permissive licence) · get_code("4d2f5f7e7d5fc1a1")
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 str. 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)

class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.linear(x)

class CustomClass:
    pass

class AnotherClass:
    pass

# Create instances to test
model_inst = SimpleModel()
custom_inst = CustomClass()
another_inst = AnotherClass()
tensor_inst = torch.tensor([1, 2, 3, 4])

# The function only calls __class__.__name__ on the instance
# So we just need any object with a __class__ attribute (all Python objects have this)
ARGS = {
    "inst": model_inst
}
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
read_file_with_assertion Ran
ran on a synthesized input
TheShadow29/VidSitu/prep_data/dwn_yt.py
code served (permissive licence) · get_code("b11ee0a6f802aabd")
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
from pathlib import Path

# Create a temporary JSON file with test data
temp_dir = tempfile.mkdtemp()
temp_file = Path(temp_dir) / "test_data.json"

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

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

ARGS = {
    "fpath": str(temp_file),
    "read_type": "r",
    "reader": "json"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — 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
read_file_with_assertion Ran
ran; our contract draft was wrong, not the code
TheShadow29/VidSitu/vidsitu_code/evl_fns.py
code served (permissive licence) · get_code("3b568ae652a0217f")
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 pickle
import numpy as np
from pathlib import Path
import json
import tempfile
import os

# Create a temporary directory for test files
temp_dir = tempfile.mkdtemp()

# Create a test JSON file
json_file_path = os.path.join(temp_dir, "test.json")
test_json_data = {"key": "value", "number": 42}
with open(json_file_path, 'w') as f:
    json.dump(test_json_data, f)

# Create a test pickle file
pickle_file_path = os.path.join(temp_dir, "test.pkl")
test_pickle_data = {"pickle_key": "pickle_value", "list": [1, 2, 3]}
with open(pickle_file_path, 'wb') as f:
    pickle.dump(test_pickle_data, f)

# Create a test numpy file
numpy_file_path = os.path.join(temp_dir, "test.npy")
test_numpy_data = np.array([1, 2, 3, 4, 5])
np.save(numpy_file_path, test_numpy_data)

ARGS = {
    'fpath': json_file_path,
    'read_type': 'r',
    'reader': 'json'
}
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 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
rem_mdl Ran
ran; our contract draft was wrong, not the code
TheShadow29/VidSitu/vidsitu_code/feat_extractor.py
code served (permissive licence) · get_code("ba87f2f7b1e31bc3")
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 str. 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

torch.manual_seed(0)

def rem_mdl(k):
    k1 = k.split('module.', 1)[1]
    return k1

ARGS = {
    'k': 'module.layer1.weight'
}
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
st_ag Ran
ran on a synthesized input
TheShadow29/VidSitu/vidsitu_code/dat_loader.py
code served (permissive licence) · get_code("7425b7ac40dc3b30")
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 str. 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

torch.manual_seed(0)

def st_ag(ag):
    return f'<{ag}>'

ARGS = {
    'ag': 'test_argument'
}
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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

Also present in other papers' code

Of 9 samples lifted from this paper's linked repositories, 9 carry a function identity and 0 do not. Of the 9 distinct functions with an identity, 6 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 6 later) and 3 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
get_name_from_inst 2 other papers
1 dated earlier, 1 later
2003.10606 (2020-03)
Video Object Grounding using Semantic Roles in Language Description
aaai_25418 (2023 (year only))
Video Event Extraction via Tracking Visual States of Arguments
collate_dct_lst_naive 1 other paper
1 later
aaai_25418 (2023 (year only))
Video Event Extraction via Tracking Visual States of Arguments
enclose_ag 1 other paper
1 later
aaai_25418 (2023 (year only))
Video Event Extraction via Tracking Visual States of Arguments
end_ag 1 other paper
1 later
aaai_25418 (2023 (year only))
Video Event Extraction via Tracking Visual States of Arguments
read_file_with_assertion 1 other paper
1 later
aaai_25418 (2023 (year only))
Video Event Extraction via Tracking Visual States of Arguments
st_ag 1 other paper
1 later
aaai_25418 (2023 (year only))
Video Event Extraction via Tracking Visual States of Arguments

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

Verb: deflect (block, avoid) Arg0 (deflector) woman with shield Arg1 (thing deflected) boulder Scene city park Verb: talk (speak) Arg0 (talker) woman with shield Arg2 (hearer) man with trident ArgM (manner) urgently Scene city park Verb: leap (physically leap) Arg0 (jumper) man with trident Arg1 (obstacle) over stairs ArgM (direction) towards shirtless man ArgM (goal) to attack shirtless man Scene city park Verb: punch (to hit) Arg0 (agent) shirtless man Arg1 (entity punched) man with trident ArgM (direction) far into distance Scene city park Verb: punch (to hit) Arg0 (agent) shirtless man Arg1 (entity punched) woman with shield ArgM (direction) down the stairs Scene city park 2 Seconds

For agents

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

get_harvested_code_for_paper("2104.00990")
get_code_for_paper("2104.00990")
have("2104.00990")

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

curl https://syntology.ai/api/ran/2104.00990.json

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

[![Syntology run record](https://syntology.ai/api/ran/2104.00990.svg)](https://syntology.ai/paper/2104.00990)

Connect an agent — have() is free.