MMTEB: Massive Multilingual Text Embedding Benchmark
Paper • 2502.13595 • Published • 50
Datasets:
id string | text string |
|---|---|
0 | a video of Accordion |
1 | a video of Acoustic guitar |
2 | a video of Baby cry, infant cry |
3 | a video of Banjo |
4 | a video of Bark |
5 | a video of Bus |
6 | a video of Cat |
7 | a video of Chainsaw |
8 | a video of Church bell |
9 | a video of Clock |
10 | a video of Female speech, woman speaking |
11 | a video of Fixed-wing aircraft, airplane |
12 | a video of Flute |
13 | a video of Frying (food) |
14 | a video of Goat |
15 | a video of Helicopter |
16 | a video of Horse |
17 | a video of Male speech, man speaking |
18 | a video of Mandolin |
19 | a video of Motorcycle |
20 | a video of Race car, auto racing |
21 | a video of Rodents, rats, mice |
22 | a video of Shofar |
23 | a video of Toilet flush |
24 | a video of Train horn |
25 | a video of Truck |
26 | a video of Ukulele |
27 | a video of Violin, fiddle |
Audio-Visual Event (AVE) classification dataset containing 28 event categories such as church bell, truck, dog barking, and other everyday sounds sourced from YouTube videos. The goal is to predict the sound event category from video only. Used the official train and test splits across 28 audio-visual event classes (~3,312 train / ~402 test).
| Task category | VideoZeroshotClassification (video-to-text) |
| Domains | AudioScene, Web |
| Reference | The European Conference on Computer Vision (ECCV) |
Source datasets:
How to evaluate on this task
You can evaluate an embedding model on this dataset using the following code:
import mteb
task = mteb.get_task("AVEDatasetVideoZeroShot")
model = mteb.get_model(YOUR_MODEL)
mteb.evaluate(model, task)
To learn more about how to run models on mteb task check out the GitHub repository.
Citation
If you use this dataset, please cite the dataset as well as mteb, as this dataset likely includes additional processing as a part of the MMTEB Contribution.
@inproceedings{Tian_2018_ECCV,
author = {Tian, Yapeng and Shi, Jing and Li, Bochen and Duan, Zhiyao and Xu, Chenliang},
booktitle = {The European Conference on Computer Vision (ECCV)},
month = {September},
title = {Audio-Visual Event Localization in Unconstrained Videos},
year = {2018},
}
@article{enevoldsen2025mmtebmassivemultilingualtext,
title={MMTEB: Massive Multilingual Text Embedding Benchmark},
author={Kenneth Enevoldsen and Isaac Chung and Imene Kerboua and Márton Kardos and Ashwin Mathur and David Stap and Jay Gala and Wissam Siblini and Dominik Krzemiński and Genta Indra Winata and Saba Sturua and Saiteja Utpala and Mathieu Ciancone and Marion Schaeffer and Gabriel Sequeira and Diganta Misra and Shreeya Dhakal and Jonathan Rystrøm and Roman Solomatin and Ömer Çağatan and Akash Kundu and Martin Bernstorff and Shitao Xiao and Akshita Sukhlecha and Bhavish Pahwa and Rafał Poświata and Kranthi Kiran GV and Shawon Ashraf and Daniel Auras and Björn Plüster and Jan Philipp Harries and Loïc Magne and Isabelle Mohr and Mariya Hendriksen and Dawei Zhu and Hippolyte Gisserot-Boukhlef and Tom Aarsen and Jan Kostkan and Konrad Wojtasik and Taemin Lee and Marek Šuppa and Crystina Zhang and Roberta Rocca and Mohammed Hamdy and Andrianos Michail and John Yang and Manuel Faysse and Aleksei Vatolin and Nandan Thakur and Manan Dey and Dipam Vasani and Pranjal Chitale and Simone Tedeschi and Nguyen Tai and Artem Snegirev and Michael Günther and Mengzhou Xia and Weijia Shi and Xing Han Lù and Jordan Clive and Gayatri Krishnakumar and Anna Maksimova and Silvan Wehrli and Maria Tikhonova and Henil Panchal and Aleksandr Abramov and Malte Ostendorff and Zheng Liu and Simon Clematide and Lester James Miranda and Alena Fenogenova and Guangyu Song and Ruqiya Bin Safi and Wen-Ding Li and Alessia Borghini and Federico Cassano and Hongjin Su and Jimmy Lin and Howard Yen and Lasse Hansen and Sara Hooker and Chenghao Xiao and Vaibhav Adlakha and Orion Weller and Siva Reddy and Niklas Muennighoff},
publisher = {arXiv},
journal={arXiv preprint arXiv:2502.13595},
year={2025},
url={https://arxiv.org/abs/2502.13595},
doi = {10.48550/arXiv.2502.13595},
}
@article{muennighoff2022mteb,
author = {Muennighoff, Niklas and Tazi, Nouamane and Magne, Loïc and Reimers, Nils},
title = {MTEB: Massive Text Embedding Benchmark},
publisher = {arXiv},
journal={arXiv preprint arXiv:2210.07316},
year = {2022}
url = {https://arxiv.org/abs/2210.07316},
doi = {10.48550/ARXIV.2210.07316},
}
Dataset Statistics
Dataset Statistics
The following code contains the descriptive statistics from the task. These can also be obtained using:
import mteb
task = mteb.get_task("AVEDatasetVideoZeroShot")
desc_stats = task.metadata.descriptive_stats
{
"test": {
"num_samples": 402,
"text_statistics": null,
"image_statistics": null,
"audio_statistics": null,
"video_statistics": {
"total_duration_seconds": 4026.542646000018,
"total_frames": 115005,
"min_width": 192,
"average_width": 769.9601990049752,
"max_width": 1280,
"min_height": 144,
"average_height": 484.2587064676617,
"max_height": 720,
"min_duration_seconds": 10.0,
"average_duration_seconds": 10.016275238806015,
"max_duration_seconds": 10.061281,
"unique_videos": 402,
"average_fps": 28.579601990049753,
"fps": {
"30": 289,
"25": 87,
"24": 18,
"27": 1,
"29": 4,
"21": 1,
"28": 1,
"20": 1
},
"min_resolution": [
192,
144
],
"average_resolution": [
769.9601990049752,
484.2587064676617
],
"max_resolution": [
1280,
720
],
"resolutions": {
"640x360": 25,
"1280x720": 143,
"480x360": 112,
"640x352": 3,
"960x720": 9,
"490x360": 1,
"480x270": 3,
"320x240": 30,
"512x288": 2,
"480x352": 1,
"540x360": 11,
"492x360": 3,
"406x720": 3,
"450x360": 6,
"640x354": 1,
"484x360": 1,
"352x288": 2,
"322x240": 1,
"360x288": 1,
"384x288": 5,
"192x144": 2,
"480x272": 2,
"352x240": 2,
"1080x720": 1,
"568x320": 3,
"640x290": 1,
"270x360": 2,
"638x360": 1,
"636x360": 1,
"426x240": 2,
"720x720": 1,
"422x360": 1,
"480x268": 1,
"480x318": 1,
"1170x720": 1,
"480x266": 1,
"202x360": 2,
"400x328": 1,
"480x320": 4,
"204x360": 3,
"198x360": 1,
"312x234": 1,
"398x238": 1,
"352x264": 1,
"480x356": 1,
"400x226": 1
}
},
"label_statistics": {
"min_labels_per_text": 1,
"average_label_per_text": 1.0,
"max_labels_per_text": 1,
"unique_labels": 28,
"labels": {
"8": {
"count": 18
},
"17": {
"count": 17
},
"4": {
"count": 17
},
"11": {
"count": 18
},
"20": {
"count": 18
},
"10": {
"count": 17
},
"15": {
"count": 17
},
"27": {
"count": 18
},
"12": {
"count": 17
},
"26": {
"count": 16
},
"13": {
"count": 18
},
"25": {
"count": 13
},
"22": {
"count": 10
},
"19": {
"count": 9
},
"1": {
"count": 18
},
"24": {
"count": 17
},
"9": {
"count": 16
},
"3": {
"count": 18
},
"14": {
"count": 10
},
"2": {
"count": 8
},
"5": {
"count": 6
},
"7": {
"count": 16
},
"6": {
"count": 6
},
"16": {
"count": 6
},
"23": {
"count": 18
},
"21": {
"count": 10
},
"0": {
"count": 15
},
"18": {
"count": 15
}
}
},
"candidates_labels_text_statistics": {
"total_text_length": 623,
"min_text_length": 14,
"average_text_length": 22.25,
"max_text_length": 40,
"unique_texts": 28
}
}
}
This dataset card was automatically generated using MTEB
- Downloads last month
- 131