SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2406.14515 · NeurIPS · 2024

MMBench-Video: A Long-Form Multi-Shot Benchmark for Holistic Video Understanding

Dahua Lin, Kai Chen, Xiangyu Zhao, Haodong Duan, Xinyu Fang, Yining Li, Kangrui Mao

arXiv · PDF · Open in the Atlas

Code that ran

5 of 6 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; 1 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 4 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/2406.14515.json · badge: /api/ran/2406.14515.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
open-compass/VLMEvalKit extension (classifier label)
classifier basis: classifier_2025
4 ran · 0 no recorded run
open-compass/vlmevalkit extension (classifier label)
classifier basis: classifier_2025
1 ran · 1 no recorded run
FunctionStatusWhere it lives
insert_subtitles_into_frames Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
open-compass/VLMEvalKit/vlmeval/dataset/longvideobench.py
code served (permissive licence) · get_code("67cb1c04a80dd0b2")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) 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 numpy as np

torch.manual_seed(0)

# Create small dummy frames (images as tensors)
frames = [
    torch.randn(3, 8, 8),  # 3 channels, 8x8 image
    torch.randn(3, 8, 8),
    torch.randn(3, 8, 8),
    torch.randn(3, 8, 8),
    torch.randn(3, 8, 8),
]

# Frame timestamps in seconds
frame_timestamps = [0.5, 1.5, 2.5, 3.5, 4.5]

# Subtitles with 'timestamp' format
subtitles = [
    {'timestamp': (1.0, 2.0), 'text': 'First subtitle'},
    {'timestamp': (3.0, 4.0), 'text': 'Second subtitle'},
    {'start': '00:00:05', 'end': '00:00:06', 'line': 'Third subtitle'},
]

# Starting timestamp offset
starting_timestamp_for_subtitles = 0.0

# Total duration
duration = 10.0

ARGS = {
    'frames': frames,
    'frame_timestamps': frame_timestamps,
    'subtitles': subtitles,
    'starting_timestamp_for_subtitles': starting_timestamp_for_subtitles,
    'duration': duration,
}
Setup block drafted by
a language model, Claude Sonnet 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 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
register_adapter Ran
ran; our contract draft was wrong, not the code
open-compass/vlmevalkit/vlmeval/api/adapters/base.py
code served (permissive licence) · get_code("c7a9d23ea67e8829")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) and returned function. 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.

from functools import partial

# The function can be used in two ways:
# 1. As a decorator (factory=None, returns decorator)
# 2. Called directly with both name and factory

# We'll test the direct call case with both name and factory provided
# since that exercises all parameters

class DummyAdapter:
    """A simple adapter class to register"""
    def __init__(self, config=None):
        self.config = config
    
    def forward(self, x):
        return x

# Create a factory function
def adapter_factory():
    return DummyAdapter(config={'param': 1})

ARGS = {
    'name': 'test_adapter',
    'factory': adapter_factory
}
Setup block drafted by
a language model, Claude Sonnet 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 function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
timestamp_to_seconds Ran
ran, honoured the contract we drafted
open-compass/VLMEvalKit/vlmeval/dataset/longvideobench.py
code served (permissive licence) · get_code("b0e798a910385a74")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) 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.

ARGS = {
    "timestamp": "01:23:45.67"
}
Setup block drafted by
a language model, Claude Sonnet 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 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
uniformly_subsample Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
open-compass/VLMEvalKit/vlmeval/dataset/longvideobench.py
code served (permissive licence) · get_code("1413da9578c40730")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) and returned list, ndim 1, dtype family int; 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 numpy as np

torch.manual_seed(0)

# The function uniformly_subsample takes a list and an integer K
# It subsamples the list to return K uniformly spaced elements

# Create a simple list to subsample
lst = [10, 20, 30, 40, 50, 60, 70, 80, 90, 100]

# K is the number of elements we want to subsample to
K = 4

ARGS = {
    "lst": lst,
    "K": K
}
Setup block drafted by
a language model, Claude Sonnet 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 list, ndim 1, dtype family int — 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
upload_file Ran
ran; our contract draft was wrong, not the code
open-compass/VLMEvalKit/scripts/mmb_eval_gradio.py
code served (permissive licence) · get_code("5bfe564b40e0899d")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) 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.

class MockFile:
    def __init__(self, name):
        self.name = name

file = MockFile("/path/to/example_file.txt")

ARGS = {
    "file": file
}
Setup block drafted by
a language model, Claude Sonnet 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 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
build_adapter Not yet run
unverified (harvested, no recorded run)
open-compass/vlmevalkit/vlmeval/api/adapters/base.py
code served (permissive licence) · get_code("ab5369905692e7e6")
How it ran

no invocation recorded.

Also present in other papers' code

Of 6 samples lifted from this paper's linked repositories, 6 carry a function identity and 0 do not. Of the 6 distinct functions with an identity, 1 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 1 later) and 5 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
timestamp_to_seconds 2 other papers
2 later
2407.15754 (2024-07)
LongVideoBench: A Benchmark for Long-context Interleaved Video-Language Understanding
2409.14485 (2024-09)
Video-XL: Extra-Long Vision Language Model for Hour-Scale Video Understanding

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

The advent of large vision-language models (LVLMs) has spurred research into their applications in multi-modal contexts, particularly in video understanding. Traditional VideoQA benchmarks, despite providing quantitative metrics, often fail to encompass the full spectrum of video content and inadequately assess models' temporal comprehension. To address these limitations, we introduce MMBench-Video, a quantitative benchmark designed to rigorously evaluate LVLMs' proficiency in video understanding. MMBench-Video incorporates lengthy videos from YouTube and employs free-form questions, mirroring practical use cases. The benchmark is meticulously crafted to probe the models' temporal reasoning skills, with all questions human-annotated according to a carefully constructed ability taxonomy. We employ GPT-4 for automated assessment, demonstrating superior accuracy and robustness over earlier LLM-based evaluations. Utilizing MMBench-Video, we have conducted comprehensive evaluations that include both proprietary and open-source LVLMs for images and videos. MMBench-Video stands as a valuable resource for the research community, facilitating improved evaluation of LVLMs and catalyzing progress in the field of video understanding. The evalutation code of MMBench-Video will be integrated into VLMEvalKit: https://github.com/open-compass/VLMEvalKit.

For agents

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

get_harvested_code_for_paper("2406.14515")
get_code_for_paper("2406.14515")
have("2406.14515")

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

curl https://syntology.ai/api/ran/2406.14515.json

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

[![Syntology run record](https://syntology.ai/api/ran/2406.14515.svg)](https://syntology.ai/paper/2406.14515)

Connect an agent — have() is free.