SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1809.01696 · 2018

TVQA: Localized, Compositional Video Question Answering

arXiv · PDF · Open in the Atlas

Code that ran

7 of 8 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; 3 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/1809.01696.json · badge: /api/ran/1809.01696.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
jayleicn/TVQA canonical (classifier label)
classifier basis: census
5 ran · 1 no recorded run
mansigoel/TVQA reimplementation (classifier label)
classifier basis: classifier_2025
2 ran · 0 no recorded run
FunctionStatusWhere it lives
get_vidname2cnt_per_show Ran
ran; our contract draft was wrong, not the code
mansigoel/TVQA/preprocessing.py
code served (permissive licence) · get_code("1b30465bb758c4b8")
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) 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 os
import tempfile
import shutil

# Create a temporary directory structure with subdirectories containing .jpg files
temp_base = tempfile.mkdtemp()

# Create subdirectories with varying numbers of .jpg files
subdir1 = os.path.join(temp_base, "video_001")
subdir2 = os.path.join(temp_base, "video_002")
subdir3 = os.path.join(temp_base, "video_003")

os.makedirs(subdir1, exist_ok=True)
os.makedirs(subdir2, exist_ok=True)
os.makedirs(subdir3, exist_ok=True)

# Create dummy .jpg files in each subdirectory
for i in range(5):
    open(os.path.join(subdir1, f"frame_{i}.jpg"), 'w').close()

for i in range(3):
    open(os.path.join(subdir2, f"frame_{i}.jpg"), 'w').close()

for i in range(7):
    open(os.path.join(subdir3, f"frame_{i}.jpg"), 'w').close()

# Also create a non-jpg file to ensure it's not counted
open(os.path.join(subdir1, "metadata.txt"), 'w').close()

ARGS = {
    "base_path": temp_base
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
tqdm — 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
load_json Ran
ran; our contract draft was wrong, not the code
jayleicn/TVQA/utils.py
code served (permissive licence) · get_code("53d425eb1278ddda")
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) 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 content
temp_dir = tempfile.mkdtemp()
temp_file = os.path.join(temp_dir, "test.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 = {
    "file_path": temp_file
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
max_along_time Ran
ran on a synthesized input
jayleicn/TVQA/model/rnn.py
code served (permissive licence) · get_code("4e8b207ee04445f8")
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, 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)

# outputs: (B, T, D) - batch size 2, time steps 5, dimension 4
outputs = torch.randn(2, 5, 4)

# lengths: (B,) - batch size 2, specifying actual sequence lengths
lengths = torch.tensor([3, 4])

ARGS = {
    "outputs": outputs,
    "lengths": lengths,
}
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, 4) — shape (2, 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
mean_along_time Ran
ran on a synthesized input
jayleicn/TVQA/model/rnn.py
code served (permissive licence) · get_code("8ec050fe9cabe7f9")
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, 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)

# outputs: (B, T, D) - batch size 2, time steps 3, feature dimension 4
outputs = torch.randn(2, 3, 4)

# lengths: (B, ) - batch size 2, specifying the actual length for each batch element
lengths = torch.tensor([2, 3])

ARGS = {
    "outputs": outputs,
    "lengths": lengths,
}
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, 4) — shape (2, 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
merge_list_dicts Ran
ran; our contract draft was wrong, not the code
mansigoel/TVQA/preprocessing.py
code served (permissive licence) · get_code("13141c69e7fed551")
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) 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)

# Analyze the function:
# - merge_list_dicts takes a list of dictionaries
# - It copies the first dict, then updates it with subsequent dicts
# - Returns the merged dictionary

# Create a list of dictionaries with distinct keys and values
dict1 = {
    'key_a': 1,
    'key_b': 'value_b',
    'tensor_1': torch.randn(2, 4)
}

dict2 = {
    'key_c': 2.5,
    'key_d': [1, 2, 3],
    'tensor_2': torch.randn(3, 4)
}

dict3 = {
    'key_e': {'nested': 'dict'},
    'key_b': 'updated_value_b'  # This will override dict1's key_b
}

list_dicts = [dict1, dict2, dict3]

ARGS = {
    'list_dicts': list_dicts
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
preprocess_inputs Ran
ran on a synthesized input
jayleicn/TVQA/tvqa_dataset.py
code served (permissive licence) · get_code("1dc3bf82cc77e497")
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. 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)

class BatchedData:
    """Stand-in class that mimics a batch object with required attributes"""
    def __init__(self):
        # Text keys: 'q', 'a0', 'a1', 'a2', 'a3', 'a4', 'sub', 'vcpt'
        # Each text key is a tuple of (tensor, length_tensor)
        batch_size = 2
        
        # q: question - no max_len constraint, just (context, lengths)
        self.q = (torch.randint(0, 100, (batch_size, 5)), torch.randint(1, 6, (batch_size,)))
        
        # a0-a4: answers - no max_len constraint
        self.a0 = (torch.randint(0, 100, (batch_size, 4)), torch.randint(1, 5, (batch_size,)))
        self.a1 = (torch.randint(0, 100, (batch_size, 4)), torch.randint(1, 5, (batch_size,)))
        self.a2 = (torch.randint(0, 100, (batch_size, 4)), torch.randint(1, 5, (batch_size,)))
        self.a3 = (torch.randint(0, 100, (batch_size, 4)), torch.randint(1, 5, (batch_size,)))
        self.a4 = (torch.randint(0, 100, (batch_size, 4)), torch.randint(1, 5, (batch_size,)))
        
        # sub: subtitles - has max_len constraint, (context, lengths)
        self.sub = (torch.randint(0, 100, (batch_size, 20)), torch.randint(1, 21, (batch_size,)))
        
        # vcpt: visual concepts - has max_len constraint, (context, lengths)
        self.vcpt = (torch.randint(0, 100, (batch_size, 15)), torch.randint(1, 16, (batch_size,)))
        
        # vid: video features - no max_len constraint, just (tensor, dummy_lengths)
        self.vid = (torch.randn(batch_size, 8, 4), torch.ones(batch_size, dtype=torch.long))
        
        # answer_idx: target labels
        self.answer_idx = torch.randint(0, 5, (batch_size,))
        
        # qid: question id
        self.qid = torch.arange(batch_size)

batched_data = BatchedData()

ARGS = {
    'batched_data': batched_data,
    'max_sub_l': 10,
    'max_vcpt_l': 8,
    'max_vid_l': 12,
    'device': 'cpu'
}
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 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
read_json_lines Ran
ran on a synthesized input
jayleicn/TVQA/utils.py
code served (permissive licence) · get_code("62891394278900f5")
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 json
import tempfile
import os

# Create a temporary JSON lines file
temp_dir = tempfile.mkdtemp()
temp_file = os.path.join(temp_dir, "test.jsonl")

# Write some test data to the file
test_data = [
    {"key1": "value1", "key2": 42},
    {"key1": "value2", "key2": 43},
    {"key1": "value3", "key2": 44}
]

with open(temp_file, 'w') as f:
    for item in test_data:
        f.write(json.dumps(item) + '\n')

ARGS = {
    "file_path": temp_file
}
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 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
pad_collate Not yet run
unverified (harvested, no recorded run)
jayleicn/TVQA/tvqa_dataset.py
code served (permissive licence) · get_code("d10981a16ea4b3bc")
How it ran

no invocation recorded.

Also present in other papers' code

Of 8 samples lifted from this paper's linked repositories, 8 carry a function identity and 0 do not. Of the 8 distinct functions with an identity, 2 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 2 later) and 6 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
load_json 11 other papers
11 later
1904.11574 (2019-04)
TVQA+: Spatio-Temporal Grounding for Video Question Answering
2001.09099 (2020-01)
TVR: A Large-Scale Dataset for Video-Subtitle Moment Retrieval
2023.findings-emnlp.644 (2023 (year only))
VIP5: Towards Multimodal Foundation Models for Recommendation
and 8 more not named here
max_along_time 1 other paper
1 later
2008.09105 (2020-08)
Location-Aware Graph Convolutional Networks for Video Question Answering

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

Recent years have witnessed an increasing interest in image-based question-answering (QA) tasks. However, due to data limitations, there has been much less work on video-based QA. In this paper, we present TVQA, a large-scale video QA dataset based on 6 popular TV shows. TVQA consists of 152,545 QA pairs from 21,793 clips, spanning over 460 hours of video. Questions are designed to be compositional in nature, requiring systems to jointly localize relevant moments within a clip, comprehend subtitle-based dialogue, and recognize relevant visual concepts. We provide analyses of this new dataset as well as several baselines and a multi-stream end-to-end trainable neural network framework for the TVQA task. The dataset is publicly available at http://tvqa.cs.unc.edu.

For agents

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

get_harvested_code_for_paper("1809.01696")
get_code_for_paper("1809.01696")
have("1809.01696")

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

curl https://syntology.ai/api/ran/1809.01696.json

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

[![Syntology run record](https://syntology.ai/api/ran/1809.01696.svg)](https://syntology.ai/paper/1809.01696)

Connect an agent — have() is free.