Pith. sign in
Pith Number

pith:T7EF4SUO

pith:2024:T7EF4SUOFTLLSQDQNLC2BWIURX
not attested not anchored not stored refs resolved

MLVU: Benchmarking Multi-task Long Video Understanding

Boya Wu, Bo Zhang, Bo Zhao, Junjie Zhou, Minghao Qin, Shitao Xiao, Tiejun Huang, Xi Yang, Yan Shu, Yongping Xiong, Zheng Liu, Zhengyang Liang

MLVU benchmark shows current multimodal models struggle with most long video tasks and degrade sharply on longer clips.

arxiv:2406.04264 v3 · 2024-06-06 · cs.CV · cs.AI · cs.CL

Add to your LaTeX paper
\usepackage{pith}
\pithnumber{T7EF4SUOFTLLSQDQNLC2BWIURX}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge

Record completeness

1 Bitcoin timestamp
2 Internet Archive
3 Author claim open · sign in to claim
4 Citations open
5 Replications open
Portable graph bundle live · download bundle · merged state
The bundle contains the canonical record plus signed events. A mirror can host it anywhere and recompute the same current state with the deterministic merge algorithm.

Claims

C1strongest claim

The empirical study with 23 latest MLLMs reveals significant room for improvement in today's technique, as all existing methods struggle with most of the evaluation tasks and exhibit severe performance degradation when handling longer videos.

C2weakest assumption

The chosen video lengths, genres, and tasks in MLVU sufficiently represent the core challenges of real-world long video understanding and that performance on these tasks generalizes beyond the benchmark.

C3one line summary

MLVU is a new benchmark for long video understanding that uses extended videos across diverse genres and multi-task evaluations, revealing that current MLLMs struggle significantly and degrade sharply with longer durations.

References

63 extracted · 63 resolved · 24 Pith anchors

[1] GPT-4 Technical Report 2023 · arXiv:2303.08774
[2] Anthropic. Claude 3. https://www.anthropic.com/ news/claude-3-family, 2024. 7, 2 2024
[3] Minigpt4-video: Advancing multimodal llms for video understanding with interleaved visual-textual tokens 2024
[4] Qwen Technical Report · arXiv:2309.16609
[5] Frozen in time: A joint video and image encoder for end-to- end retrieval 2021

Formal links

2 machine-checked theorem links

Cited by

64 papers in Pith

Receipt and verification
First computed 2026-05-17T23:39:21.958186Z
Builder pith-number-builder-2026-05-17-v1
Signature Pith Ed25519 (pith-v1-2026-05) · public key
Schema pith-number/v1.0

Canonical hash

9fc85e4a8e2cd6b940706ac5a0d9148de724540951e474f6b45f20ab24059c3a

Aliases

arxiv: 2406.04264 · arxiv_version: 2406.04264v3 · doi: 10.48550/arxiv.2406.04264 · pith_short_12: T7EF4SUOFTLL · pith_short_16: T7EF4SUOFTLLSQDQ · pith_short_8: T7EF4SUO
Agent API
Verify this Pith Number yourself
curl -sH 'Accept: application/ld+json' https://pith.science/pith/T7EF4SUOFTLLSQDQNLC2BWIURX \
  | jq -c '.canonical_record' \
  | python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: 9fc85e4a8e2cd6b940706ac5a0d9148de724540951e474f6b45f20ab24059c3a
Canonical record JSON
{
  "metadata": {
    "abstract_canon_sha256": "41148e2d78f4d31b4e397c1a844d4b6f5ce24671e5f8064a2655bd442342726a",
    "cross_cats_sorted": [
      "cs.AI",
      "cs.CL"
    ],
    "license": "http://creativecommons.org/licenses/by-nc-sa/4.0/",
    "primary_cat": "cs.CV",
    "submitted_at": "2024-06-06T17:09:32Z",
    "title_canon_sha256": "18a5bf14850be2ac26f321769b1dc07c5948b5228a893e1a555f8f75b88e6ccd"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2406.04264",
    "kind": "arxiv",
    "version": 3
  }
}