README.md

---
license: mit
task_categories:
- multiple-choice
- audio-classification
- visual-question-answering
language:
- en
- zh
tags:
- audio-visual
- omni-modality
- cross-modal
- consistency
- benchmark
pretty_name: XModBench
size_categories:
- 10K<n<100K
configs:
- config_name: audio_text
  data_files: data/audio_text.jsonl
- config_name: text_audio
  data_files: data/text_audio.jsonl
- config_name: audio_image
  data_files: data/audio_image.jsonl
- config_name: image_audio
  data_files: data/image_audio.jsonl
- config_name: image_text
  data_files: data/image_text.jsonl
- config_name: text_image
  data_files: data/text_image.jsonl
- config_name: audio_video
  data_files: data/audio_video.jsonl
- config_name: text_video
  data_files: data/text_video.jsonl
- config_name: video_audio
  data_files: data/video_audio.jsonl
- config_name: video_text
  data_files: data/video_text.jsonl
- config_name: lite_a2t
  data_files: data_lite/a2t.jsonl
- config_name: lite_a2v
  data_files: data_lite/a2v.jsonl
- config_name: lite_t2a
  data_files: data_lite/t2a.jsonl
- config_name: lite_t2v
  data_files: data_lite/t2v.jsonl
- config_name: lite_v2a
  data_files: data_lite/v2a.jsonl
- config_name: lite_v2t
  data_files: data_lite/v2t.jsonl
---

<h1 align="center">XModBench</h1>

<p align="center">
  <b>Benchmarking Cross-Modal Capabilities and Consistency in Omni-Language Models</b>
</p>

<p align="center">
  <a href="https://iclr.cc/Conferences/2026"><img src="https://img.shields.io/badge/ICLR-2026-8e44ad.svg" alt="ICLR 2026"></a>
  <a href="https://arxiv.org/abs/2510.15148"><img src="https://img.shields.io/badge/arXiv-2510.15148-b31b1b.svg" alt="Paper"></a>
  <a href="https://xingruiwang.github.io/projects/XModBench/"><img src="https://img.shields.io/badge/Website-Page-0a7aca?logo=globe&logoColor=white" alt="Website"></a>
  <a href="https://github.com/XingruiWang/XModBench"><img src="https://img.shields.io/badge/GitHub-Code-181717?logo=github&logoColor=white" alt="GitHub"></a>
  <a href="https://github.com/XingruiWang/lmms-eval"><img src="https://img.shields.io/badge/lmms--eval-Integration-4b9cd3.svg" alt="lmms-eval"></a>
  <a href="https://opensource.org/licenses/MIT"><img src="https://img.shields.io/badge/License-MIT-green.svg" alt="License: MIT"></a>
</p>

<p align="center">
  <img src="https://xingruiwang.github.io/projects/XModBench/static/images/teaser.png" width="92%" alt="XModBench teaser">
</p>

<p align="center"><i>🎉 Accepted at <b>ICLR 2026</b></i></p>

## What is XModBench?

**XModBench** is the first tri-modal (audio / vision / text) multiple-choice
QA benchmark explicitly designed to measure **cross-modal consistency** — does
an omni-language model give the same correct answer when the *same* semantic
content is presented in different modalities?

Each item is a 4-choice question with a `<context>` (question stem) and four
`<candidates>` (options). By permuting which modality carries the context vs.
the candidates, every question is instantiated in **six modality
configurations**, so no single modality is privileged.

| | |
|---|---|
| **Samples** | 61,320 QA pairs |
| **Task families** | 5 — Perception, Spatial, Temporal, Linguistic, Knowledge |
| **Subtasks** | 17 |
| **Modality configs** | 6 — A→T, A→V, T→A, T→V, V→A, V→T |
| **Lite split** | 6,000 — balanced 5 families × 6 configs × 200 |
| **Languages** | English, Chinese (speech translation) |

## Repository layout

```
RyanWW/XModBench/
├── data/            # 10 JSONL files, one per raw modality combination
│   ├── audio_text.jsonl  text_audio.jsonl  audio_image.jsonl  ...
├── data_lite/       # 6 JSONL — XModBench-Lite (a2t,a2v,t2a,t2v,v2a,v2t)
├── Data.zip         # ALL media (audio/image/video) — download + unzip → Data/
├── tasks/           # original per-subtask task definitions (JSON)
└── eval_logs/       # released per-sample model outputs (reproduced via lmms-eval)
    └── <model>/<lite|full>/  samples_*.jsonl + summary.json
```

> **Media live in `Data.zip`.** The JSONL question files (`data/`,
> `data_lite/`) reference media by repo-relative paths like
> `Data/vggss_audio_bench/xxx.wav`. Download and unzip `Data.zip` once so
> those paths resolve. (`Data.zip` was rebuilt with Chapter-stripped
> `emotions/` clips — a fix for a moviepy parsing crash; see Changelog.)

## Loading the data

**1. Get the media** (one-time, ~30 GB):

```bash
huggingface-cli download RyanWW/XModBench Data.zip \
    --repo-type dataset --local-dir .
unzip Data.zip          # → ./Data/...   (matches the JSONL paths)
```

**2. Load the questions**:

```python
from datasets import load_dataset

# one modality configuration (full set)
ds = load_dataset("RyanWW/XModBench", "audio_text", split="train")

# XModBench-Lite (balanced 6k)
lite = load_dataset("RyanWW/XModBench", "lite_a2t", split="train")

# media path for the first item (resolve against the unzipped Data/)
print(ds[0]["conditions"]["input"])   # e.g. Data/vggss_audio_bench/....wav
```

The [lmms-eval port](https://github.com/XingruiWang/lmms-eval) handles the
download + path resolution automatically — no manual unzip needed there.

### Sample schema

```json
{
  "index": 1,
  "subtask": "01_perception/finegrained",
  "question": "Listen to this audio clip. Which text description best matches the sound you hear? Answer with A, B, C, or D",
  "conditions": { "modality": "Audio", "input": "Data/vggss_audio_bench/ymuNh7Cwhrs_000040.wav" },
  "options": {
    "A": { "modality": "Text", "input": "dog howling" },
    "B": { "modality": "Text", "input": "chicken clucking" },
    "C": { "modality": "Text", "input": "alligators, crocodiles hissing" },
    "D": { "modality": "Text", "input": "cuckoo bird calling" }
  },
  "correct_answer": "A",
  "category": "Animal Sounds"
}
```

- `conditions.input` / `options[*].input` are **repo-relative media paths**
  (`Data/...`) for non-text modalities, or the literal text for `Text`.
- `correct_answer` ∈ {A, B, C, D}; `subtask` is `NN_family/subtask`.

## Modality configurations

| Code | Context → Candidates |
|------|----------------------|
| A→T | Audio → Text |
| A→V | Audio → Vision (image/video) |
| T→A | Text → Audio |
| T→V | Text → Vision |
| V→A | Vision → Audio |
| V→T | Vision → Text |

`data/` keeps Image and Video separate (10 files) for efficient loading;
`data_lite/` merges Vision = Image ∪ Video into the 6 canonical configs.

## XModBench-Lite

A 6,000-sample split, **balanced** across 5 task families × 6 configs × 200,
for fast, low-cost evaluation. It tracks full-set model rankings closely
(see leaderboard) and is the recommended quick-eval target.

## Evaluate with lmms-eval

XModBench is pre-integrated in
[**XingruiWang/lmms-eval**](https://github.com/XingruiWang/lmms-eval); the
dataset auto-downloads on first run.

```bash
git clone https://github.com/XingruiWang/lmms-eval.git
cd lmms-eval && pip install -e ".[all]"

# XModBench-Lite, all 6 configs (resource-aware GPU profile)
./submit_lite.sh qwen2_5_omni_interleave Qwen/Qwen2.5-Omni-7B qwenomni3

# Level-2 metrics: by-config / by-family / disparity / imbalance
python lmms_eval/tasks/xmod_bench/summarize.py \
    --logs logs/xmod_bench_lite/results_qwen2_5_omni_interleave/
```

Per-sample model outputs we reproduced are released here under
[`eval_logs/`](https://huggingface.co/datasets/RyanWW/XModBench/tree/main/eval_logs).

## Leaderboard — XModBench-Lite (reproduced via lmms-eval)

By-config accuracy (%); **Avg.** is the mean over the six configs.

| Model | A→T | A→V | T→A | T→V | V→A | V→T | Avg. |
|-------|----:|----:|----:|----:|----:|----:|-----:|
| Qwen3-Omni-30B | 71.6 | 52.0 | 62.5 | 67.0 | 55.6 | 83.1 | **65.3** |
| Qwen2.5-Omni-7B | 63.1 | 49.8 | 59.2 | 62.5 | 50.3 | 76.4 | 60.2 |
| Baichuan-Omni-1.5 | 52.5 | 32.0 | 47.6 | 56.6 | 47.0 | 77.7 | 52.2 |
| OmniVinci | 62.2 | — | — | — | — | 78.8 | — |

Qwen2.5-Omni matches its full-set paper numbers within 5 points on every
configuration. Full-set numbers for all 14 paper models are on the
[project website](https://xingruiwang.github.io/projects/XModBench/#leaderboard).

## Changelog

- **2026-05**: `Data.zip` rebuilt — the `emotions/` MELD clips had MP4
  *Chapter* metadata that crashed `moviepy`'s parser (used by some
  evaluation backends). All emotion clips were re-muxed with
  `ffmpeg -map_chapters -1` (video/audio streams untouched). Frame content
  is identical; only the Chapter atom was removed. No other media changed.

## License

Released under the **MIT License**. Media are redistributed for research use;
please also respect the licenses of the underlying source datasets
(VGG-Sound, STARSS23, GTZAN, URMP, MELD, URBANSAS, and others).

## Citation

```bibtex
@inproceedings{wang2026xmodbench,
  title     = {XModBench: Benchmarking Cross-Modal Capabilities and Consistency in Omni-Language Models},
  author    = {Wang, Xingrui and Liu, Jiang and Huang, Chao and Yu, Xiaodong and Wang, Ze and Sun, Ximeng and Wu, Jialian and Yuille, Alan and Barsoum, Emad and Liu, Zicheng},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year      = {2026},
  url       = {https://arxiv.org/abs/2510.15148}
}
```