Initial commit

Co-authored-by: Jay Huang <Jay Huang@users.noreply.huggingface.co>

.gitattributes

@@ -0,0 +1,142 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bin.* filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.mat filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.hdf5 filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+ 
+*.zstandard filter=lfs diff=lfs merge=lfs -text
+*.tfevents* filter=lfs diff=lfs merge=lfs -text
+*.db* filter=lfs diff=lfs merge=lfs -text
+*.ark* filter=lfs diff=lfs merge=lfs -text
+**/*ckpt*data* filter=lfs diff=lfs merge=lfs -text
+**/*ckpt*.meta filter=lfs diff=lfs merge=lfs -text
+**/*ckpt*.index filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.jpg filter=lfs diff=lfs merge=lfs -text
+ 
+*.jpeg filter=lfs diff=lfs merge=lfs -text
+*.bmp filter=lfs diff=lfs merge=lfs -text
+*.gif filter=lfs diff=lfs merge=lfs -text
+*.webp filter=lfs diff=lfs merge=lfs -text
+*.mp3 filter=lfs diff=lfs merge=lfs -text
+*.wav filter=lfs diff=lfs merge=lfs -text
+*.wma filter=lfs diff=lfs merge=lfs -text
+*.aac filter=lfs diff=lfs merge=lfs -text
+*.ogg filter=lfs diff=lfs merge=lfs -text
+*.m4a filter=lfs diff=lfs merge=lfs -text
+*.m3u8 filter=lfs diff=lfs merge=lfs -text
+*.amr filter=lfs diff=lfs merge=lfs -text
+*.audio filter=lfs diff=lfs merge=lfs -text
+*.avi filter=lfs diff=lfs merge=lfs -text
+*.flv filter=lfs diff=lfs merge=lfs -text
+*.mp4 filter=lfs diff=lfs merge=lfs -text
+*.mpg filter=lfs diff=lfs merge=lfs -text
+*.asf filter=lfs diff=lfs merge=lfs -text
+*.mov filter=lfs diff=lfs merge=lfs -text
+*.mpeg filter=lfs diff=lfs merge=lfs -text
+*.3gp filter=lfs diff=lfs merge=lfs -text
+*.wmv filter=lfs diff=lfs merge=lfs -text
+*.rmvb filter=lfs diff=lfs merge=lfs -text
+*.rm filter=lfs diff=lfs merge=lfs -text
+ 
+*.mkv filter=lfs diff=lfs merge=lfs -text
+*.flash filter=lfs diff=lfs merge=lfs -text
+*.vob filter=lfs diff=lfs merge=lfs -text
+*.pdf filter=lfs diff=lfs merge=lfs -text
+*.ost filter=lfs diff=lfs merge=lfs -text
+*.pst filter=lfs diff=lfs merge=lfs -text
+*.doc filter=lfs diff=lfs merge=lfs -text
+*.docx filter=lfs diff=lfs merge=lfs -text
+*.txt filter=lfs diff=lfs merge=lfs -text
+*.ppt filter=lfs diff=lfs merge=lfs -text
+*.pptx filter=lfs diff=lfs merge=lfs -text
+*.xls filter=lfs diff=lfs merge=lfs -text
+*.xlsx filter=lfs diff=lfs merge=lfs -text
+*.vsd filter=lfs diff=lfs merge=lfs -text
+*.vsdx filter=lfs diff=lfs merge=lfs -text
+ 
+ 
+dataset_infos.json ignore
+ 
+*.tsv filter=lfs diff=lfs merge=lfs -text
+
+questions/all_questions.jsonl filter=lfs diff=lfs merge=lfs -text
+questions/by_family/IE.jsonl filter=lfs diff=lfs merge=lfs -text
+nback/action_cases.jsonl filter=lfs diff=lfs merge=lfs -text
+questions/by_family/SM.jsonl filter=lfs diff=lfs merge=lfs -text
+questions/by_family/FSA.jsonl filter=lfs diff=lfs merge=lfs -text
+questions/non_nback_questions.jsonl filter=lfs diff=lfs merge=lfs -text
+questions/by_family/DA.jsonl filter=lfs diff=lfs merge=lfs -text
+questions/nback_questions.jsonl filter=lfs diff=lfs merge=lfs -text
+nback/scene_cases.jsonl filter=lfs diff=lfs merge=lfs -text
+questions/by_family/MI.jsonl filter=lfs diff=lfs merge=lfs -text
+
+archives/non_nback_memory_interference.zip filter=lfs diff=lfs merge=lfs -text
+
+archives/nback_selected_segments.zip filter=lfs diff=lfs merge=lfs -text
+
+archives/non_nback_memory_interference.z01 filter=lfs diff=lfs merge=lfs -text
+
+archives/non_nback_memory_interference.z02 filter=lfs diff=lfs merge=lfs -text
+
+archives/non_nback_interleaved.z01 filter=lfs diff=lfs merge=lfs -text
+
+archives/non_nback_interleaved.z02 filter=lfs diff=lfs merge=lfs -text
+
+archives/non_nback_interleaved.z03 filter=lfs diff=lfs merge=lfs -text
+
+archives/non_nback_interleaved.zip filter=lfs diff=lfs merge=lfs -text
+
+archives/non_nback_split_screen.z01 filter=lfs diff=lfs merge=lfs -text
+
+archives/non_nback_split_screen.z02 filter=lfs diff=lfs merge=lfs -text
+
+archives/non_nback_split_screen.z03 filter=lfs diff=lfs merge=lfs -text
+
+archives/non_nback_split_screen.z04 filter=lfs diff=lfs merge=lfs -text
+
+archives/non_nback_split_screen.zip filter=lfs diff=lfs merge=lfs -text
+
+
+archives/nback_prefix_videos.zip filter=lfs diff=lfs merge=lfs -text
+
+figures/readme/divided-attention-example.png filter=lfs diff=lfs merge=lfs -text
+
+figures/readme/interleaved-events-example.png filter=lfs diff=lfs merge=lfs -text
+figures/readme/divided-attention-result.png filter=lfs diff=lfs merge=lfs -text
+figures/readme/interleaved-events-result.png filter=lfs diff=lfs merge=lfs -text
+
+figures/readme/memory-interference-result.png filter=lfs diff=lfs merge=lfs -text
+figures/readme/memory-interference-example.png filter=lfs diff=lfs merge=lfs -text
+figures/readme/n-back-example.png filter=lfs diff=lfs merge=lfs -text
+
+figures/readme/overview.png filter=lfs diff=lfs merge=lfs -text
+figures/readme/n-back-result.png filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,221 @@
+---
+pretty_name: M³Eval
+language:
+- en
+size_categories:
+- 1K<n<10K
+task_categories:
+- visual-question-answering
+- multiple-choice
+tags:
+- video
+- multimodal
+- benchmark
+- memory
+- long-video
+---
+<h1 align="center">M³Eval: Multi-Modal Memory Evaluation through Cognitively-Grounded Video Tasks</h1>
+
+<p align="center">
+  <a href="https://arxiv.org/search/?query=M3Eval%3A+Multi-Modal+Memory+Evaluation+through+Cognitively-Grounded+Video+Tasks&searchtype=all&source=header"><img src="https://img.shields.io/badge/Paper-arXiv-b31b1b" alt="Paper"></a>
+  <a href="https://pku-value-lab.github.io/m3eval-homepage/"><img src="https://img.shields.io/badge/Project-Page-blue" alt="Project Page"></a>
+  <a href="https://huggingface.co/datasets/PKU-VaLuE-Lab/m3eval"><img src="https://img.shields.io/badge/Hugging%20Face-Dataset-yellow" alt="Hugging Face Dataset"></a>
+  <a href="https://github.com/PKU-VaLuE-Lab/m3eval"><img src="https://img.shields.io/badge/GitHub-Code-black" alt="GitHub Code"></a>
+</p>
+
+<p align="center">
+  <a href="https://jie-1203.github.io/">Jie Huang</a><sup>1,*</sup>&emsp;
+  <a href="https://liuruixun.github.io/">Ruixun Liu</a><sup>1,*</sup>&emsp;
+  <a href="https://github.com/PKU-VaLuE-Lab/m3eval">Sirui Sun</a><sup>1</sup>&emsp;
+  <a href="https://yyxxyy574.github.io/xinyiyang.github.io/">Xinyi Yang</a><sup>1</sup>&emsp;
+  <a href="https://yinli-vision.github.io/">Yin Li</a><sup>2</sup>&emsp;
+  <a href="https://yzhu.io/">Yixin Zhu</a><sup>1</sup>&emsp;
+  <a href="https://yiwuzhong.notion.site/homepage">Yiwu Zhong</a><sup>1,†</sup>
+</p>
+
+<p align="center">
+  <sup>1</sup>Peking University&emsp;
+  <sup>2</sup>University of Wisconsin-Madison
+</p>
+
+<p align="center">
+  * Equal contribution.&emsp;† Corresponding author.
+</p>
+
+## News
+
+- **2026-6-4**: We released the M³Eval benchmark, code, and project page.
+
+## M³Eval Overview
+
+<p align="center">
+  <img src="figures/readme/overview.png" alt="M³Eval overview" width="95%">
+</p>
+
+### Abstract
+
+As multi-modal models advance towards long-form video understanding, memory emerges as a critical capability. Despite substantial effort in developing video datasets and benchmarks, existing work primarily focuses on perception and reasoning, without systematically evaluating memory: what models retain, how faithfully information is preserved, and how robust memory remains under interference.
+
+To address this gap, we introduce M³Eval, the first comprehensive evaluation framework and benchmark for probing different memory dimensions in multi-modal models. Grounded in cognitive psychology, our design features carefully constructed tasks isolating key aspects of memory. Leveraging M³Eval, we conduct extensive experiments across representative multi-modal models, revealing consistent weaknesses and distinctive behaviors.
+
+We find that models struggle to maintain disentangled representations when processing parallel video streams, exhibit interference patterns differing substantially from those observed in human memory, ground memory sources more reliably in the spatial domain than the temporal domain, and demonstrate limited symbolic memory. Collectively, our benchmark provides a valuable resource for future research, whereas our findings highlight memory as a fundamental yet underexplored capability and offer insights for designing more effective memory mechanisms in multi-modal models.
+
+## Main Results
+
+### Divided Attention
+
+![Divided Attention main result](figures/readme/divided-attention-result.png)
+
+Accuracy (%) on three divided attention metrics under the split-screen setting without swaps and with frequent left/right swaps.
+
+### Memory Interference
+
+![Memory Interference main result](figures/readme/memory-interference-result.png)
+
+Proactive: first video (V1) interferes with recall of the second video (V2); retroactive: second video (V2) interferes with recall of the first video (V1). Delta denotes proactive minus retroactive.
+
+### Interleaved Events
+
+![Interleaved Events main result](figures/readme/interleaved-events-result.png)
+
+Accuracy (%) on four interleaved reconstruction metrics.
+
+### N-Back
+
+![N-Back main result](figures/readme/n-back-result.png)
+
+Average accuracy under two symbolic attributes, scene and action, averaged over all K and N configurations.
+
+## Usage
+
+### Install
+
+```bash
+git clone https://github.com/PKU-VaLuE-Lab/m3eval.git
+cd m3eval/lmms-eval
+uv pip install -e ".[all]"
+cd ..
+```
+
+### Download the Dataset
+
+The benchmark is composed from existing public benchmarks. Users must follow the original licenses and usage terms of each source dataset.
+
+Download the dataset and unpack it into `data/m3eval/`:
+
+```bash
+huggingface-cli download PKU-VaLuE-Lab/m3eval \
+  --repo-type dataset \
+  --local-dir data/m3eval
+
+bash data/m3eval/unpack_archives.sh
+```
+
+After unpacking, `data/m3eval/` should contain:
+
+```text
+data/m3eval/
+├── qa_root/
+├── questions/
+├── nback/
+└── videos/
+    ├── interleaved/
+    ��── memory_interference/
+    ├── split_screen/
+    └── nback/
+```
+
+### Evaluation
+
+The main public entry point is:
+
+```bash
+bash lmms-eval/scripts/run_m3eval_vllm.sh \
+  --model_path /path/to/your/model \
+  --task m3eval \
+  --gpus 0 \
+  --batch_size 1
+```
+
+For multi-GPU data-parallel evaluation, use:
+
+```bash
+bash lmms-eval/scripts/run_m3eval_sharded_vllm.sh \
+  --model_path /path/to/your/model \
+  --task m3eval \
+  --gpus 0,1,2,3 \
+  --num_processes 4 \
+  --batch_size 1
+```
+
+For a quick smoke run:
+
+```bash
+bash lmms-eval/scripts/run_m3eval_vllm.sh \
+  --model_path /path/to/your/model \
+  --task m3eval_memory_interference \
+  --gpus 0 \
+  --limit 1 \
+  --max_frame_num 8
+```
+
+Useful task names:
+
+- `m3eval`
+- `m3eval_memory_interference`
+- `m3eval_split_screen`
+- `m3eval_interleaved`
+- `m3eval_nback`
+
+The script writes outputs to `lmms-eval/output/m3eval/` by default. See [lmms-eval/README_M3EVAL.md](lmms-eval/README_M3EVAL.md) for the dataset layout and task names.
+
+To convert lmms-eval outputs into paper-facing tables:
+
+```bash
+python lmms-eval/scripts/aggregate_m3eval_results.py \
+  lmms-eval/output/m3eval/path/to/*_results.json
+```
+
+## Dataset Examples
+
+### Divided Attention
+
+![Divided Attention dataset example](figures/readme/divided-attention-example.png)
+
+Simultaneous memory for two side-by-side videos.
+
+<details>
+<summary>Click to expand more examples</summary>
+
+### Memory Interference
+
+![Memory Interference dataset example](figures/readme/memory-interference-example.png)
+
+Interference between sequentially presented videos.
+
+### Interleaved Events
+
+![Interleaved Events dataset example](figures/readme/interleaved-events-example.png)
+
+Memory reconstruction from temporally interleaved clips.
+
+### N-Back
+
+![N-Back dataset example](figures/readme/n-back-example.png)
+
+Decide whether the final clip matches the clip N positions earlier.
+
+</details>
+
+## Citation
+
+If you use M³Eval in your work, please cite:
+
+```bibtex
+@article{huang2026m3eval,
+  title   = {M3Eval: Multi-Modal Memory Evaluation through Cognitively-Grounded Video Tasks},
+  author  = {Huang, Jie and Liu, Ruixun and Sun, Sirui and Yang, Xinyi and Li, Yin and Zhu, Yixin and Zhong, Yiwu},
+  journal = {arXiv preprint arXiv:XXXX.XXXXX},
+  year    = {2026}
+}
+```

archives/nback_prefix_videos.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1493a0856c444eba616d81edeb342f4f4a3b0ea4691c7cf51a1c00156fb6caa6
+size 58953003685

archives/non_nback_interleaved.z01

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6a84ff49b475205d8d8b584af9f32e501fffea3caea3710fdcd43828160a5015
+size 48318382080

archives/non_nback_interleaved.z02

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:227d50d6b7cc70637aa4422ded747f3be5391513000ae28252f04ad4db5f7967
+size 48318382080

archives/non_nback_interleaved.z03

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:761162ab831fe0db97bea38e102e544b285af66ceef826598f6247e822f37ce2
+size 48318382080

archives/non_nback_interleaved.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9795442d8dad8ee00b56b0f1704db828569dca92760d5b7a0bc57d7a2ad92f40
+size 20661718769

archives/non_nback_memory_interference.z01

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8e5f85c45e756e8f5c17b034af30309ff9a2899d8ea269dd70c417075dec0dc8
+size 48318382080

archives/non_nback_memory_interference.z02

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2ec6b9e88061e54410f94332369c0cbf3f34e51f837d54b616d04f965c4dc3de
+size 48318382080

archives/non_nback_memory_interference.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:755d7b3c2b811ba889c423cc6bc7dbd3aa041b0f7bb019415cb110c6e2c35357
+size 41523007717

archives/non_nback_split_screen.z01

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:941b3c2cfdda10a7c05cb93a3114cb40580c6837097bf66f0905c697d166e7e8
+size 48318382080

archives/non_nback_split_screen.z02

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c1487948c1c9b7263a8408d90a3df8208485777ef8c3cea253dda5f953aee9d1
+size 48318382080

archives/non_nback_split_screen.z03

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a315ffd8f25479f2a731bd7c6e6ccfbc0b682037bc4e11b0b0352cddc3bd5bb1
+size 48318382080

archives/non_nback_split_screen.z04

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f80ba0bb22b7d806a4aed5bc1366bae3a557a17ce3aff1b36d440247fb5bc1b6
+size 48318382080

archives/non_nback_split_screen.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64e5d168e87b0a711b05ad8a912b3065b16196fa08f253a2f9e5d92bf05e2825
+size 34467833253

checksums_sha256.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:66d7aeb833e63afad645347b35bb29d4b6db7fcc57fc06d11eadba20dcc35484
+size 2463

nback/action_cases.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eeeec4129909ebbe929ffcef601cf395382881718c79013cf27f7785e3ad8a6f
+size 68139

nback/scene_cases.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:02814ad64627c05837d603f184b966622378e805f473216bb8f0bde8b686dc54
+size 65302

questions/all_questions.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:55d7f18e274fdf03ec41c2798fc0d2e5c1c03f9cc9db75c114f000263a33d4ce
+size 7296430

questions/by_family/DA.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f4e52d1673bf5c5a84fc1f372c590c9dca61d382378e4ff245c8e9f78d4d9578
+size 928282

questions/by_family/FSA.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d8f47b2e4e87bc8b82fd967028b2e65accee9831334177fbfb8ca3fb3a11e335
+size 160681

questions/by_family/IE.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6c923b1bb1fd32a4ca10bc7cb59ed3f53ba0bea72c7fff5bf82c0b6b34c6f902
+size 571645

questions/by_family/MI.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:494078f330e28ae2b22e200a1b950e3d0abc86aed428095d20e9a0d3ecc714e9
+size 175102

questions/by_family/SM.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:714615e562cb2320e57f0dfab4790ff74558abedd93271d801da819a7e7ddd99
+size 134788

questions/nback_questions.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e9601aeb9bef4ba804b989f0f03604fc88a653dbc2226a98160238cce14c0be6
+size 5325932

questions/non_nback_questions.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0598d75a8d255627031437227c882852f8cc4a1b5b9b7adfa845e0289d3aa0e7
+size 1970498

unpack_archives.sh

@@ -0,0 +1,33 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+ROOT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+ARCHIVE_DIR="${ROOT_DIR}/archives"
+
+for archive in "${ARCHIVE_DIR}"/*.zip; do
+  [ -e "${archive}" ] || continue
+  echo "[unpack] ${archive}"
+  split_head="${archive%.zip}.z01"
+  if [ -f "${split_head}" ]; then
+    if command -v 7z >/dev/null 2>&1; then
+      7z x "${archive}" -o"${ROOT_DIR}"
+      continue
+    fi
+    if command -v zip >/dev/null 2>&1 && command -v unzip >/dev/null 2>&1; then
+      merged_zip="${ROOT_DIR}/.$(basename "${archive%.zip}")_merged.zip"
+      rm -f "${merged_zip}"
+      zip "${archive}" -s 0 --out "${merged_zip}"
+      unzip -oq "${merged_zip}" -d "${ROOT_DIR}"
+      rm -f "${merged_zip}"
+      continue
+    fi
+    echo "[skip] need 7z or zip+unzip for split archive ${archive}" >&2
+    continue
+  fi
+
+  if command -v unzip >/dev/null 2>&1; then
+    unzip -oq "${archive}" -d "${ROOT_DIR}"
+  else
+    python3 -m zipfile -e "${archive}" "${ROOT_DIR}"
+  fi
+done