---
title: "Cross-lingual Matryoshka Representation Learning across"
description: "They build what they describe as the first bilingual speech-text Matryoshka embedding model for Wolof and French. 31 references."
authors: ["Yaya SY", "Dioula Doucouré", "Christophe Cerisara", "Irina Illina"]
year: 2026
venue: "Findings of the Association for Computational Linguistics: ACL 2026"
source: "https://aclanthology.org/2026.findings-acl.1710/"
doi: "10.18653/v1/2026.findings-acl.1710"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/sy-2026-cross-lingual-matryoshka-representation-learning/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/sy-2026-cross-lingual-matryoshka-representation-learning/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/11/) / Cross-lingual Matryoshka Representation Learning across Speech and Text

# Cross-lingual Matryoshka Representation Learning across Speech and Text

**Authors:** Yaya SY, Dioula Doucouré, [Christophe Cerisara](https://md.research.elamur.ai/free/en/authors/christophe-cerisara/) and Irina Illina · **Year:** 2026 · **Venue:** [Findings of the Association for Computational Linguistics: ACL 2026](https://md.research.elamur.ai/free/en/venues/findings/) ([Findings of the Association for Computational Linguistics 2026](https://md.research.elamur.ai/free/en/venues/findings/2026/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.findings-acl.1710/) · [PDF](https://aclanthology.org/2026.findings-acl.1710.pdf) · [DOI 10.18653/v1/2026.findings-acl.1710](https://doi.org/10.18653/v1/2026.findings-acl.1710) · [OpenAlex](https://openalex.org/W7166881802)

**Citations:** 1 citations (OpenAlex, 2026-10-14)

## Summary
Many Wolof speakers in Senegal rely mainly on speech, but much of the administrative and educational content they need exists only as French text. The authors address both the language gap and the modality gap. They build what they describe as the first bilingual speech-text Matryoshka embedding model for Wolof and French. It lets spoken Wolof queries retrieve French documents directly, avoiding a costly ASR-plus-translation cascade that compounds errors.

For training, they assemble:
- text query-document pairs, both translated and synthetic;
- roughly 860 hours of cleaned, spontaneous Wolof speech, whose transcripts are used to generate synthetic French documents.

They also release two new benchmarks. Starting from a Matryoshka text embedder fine-tuned on Wolof-French, they compare a CLIP-style dual encoder against late fusion of HuBERT speech features into the frozen text model. Late fusion wins consistently. The model also transfers to tasks it was not trained on, such as spoken intent detection.

A rank analysis shows that information sits in a small number of components. This suggests Matryoshka compression is not yet optimal, which matters for cheap deployment in low-resource, mostly oral language communities.

## Key points
- **Speech data:** 1.4 TB of scraped Wolof podcasts and radio were filtered to 860 hours of 3–30 s utterances with DNSMOS > 3.2. The filtering steps were source separation, pyannote diarization and Silero-VAD. About a quarter of the transcripts survived perplexity and lexical-diversity filtering. They were then translated and used to generate French stories, dialogues and blogposts with Gemini-2.5-Flash.
- **Text data:** 1,176,908 Wolof-query / French-document pairs, totalling about 593M document tokens. Sources were French mMARCO, Senegalese French webpages, French QA datasets and Wolof-French translation pairs.
- **Benchmarks:**
  - Fleurs-Retrieval-Eval: 166 Wolof speech / French web document pairs, built from SIB-Fleurs.
  - Kallaama-Retrieval-Eval: 150 natural conversational queries, each paired with 3 French documents.
- **Models and training:**
  - Qwen3-0.6-Embedding, trained with InfoNCE and in-batch negatives at Matryoshka dimensions 128/256/512/1024.
  - A Wolof HuBERT speech encoder, combined via late fusion (CNN downsampling ×2 plus a projection) or via a dual encoder with attention pooling.
  - Instruction prompts select among document retrieval, speech-translation retrieval and transcription retrieval.
- **Analysis:**
  - Cumulative energy analysis shows that higher Matryoshka dimensions are low-rank, yet smaller slices still underperform.
  - A chromadb cost study shows diminishing accuracy returns relative to growing disk use.
  - Code, models and data are to be released under a non-commercial research license.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

## Institutions

- [Centre National de la Recherche Scientifique](https://md.research.elamur.ai/free/en/labs/centre-national-de-la-recherche-scientifique/) — Research institute, Paris, France · 169 works · 62,488 citations
- [Laboratoire Lorrain de Recherche en Informatique et ses Applications](https://md.research.elamur.ai/free/en/labs/laboratoire-lorrain-de-recherche-en-informatique-et-ses-applications/) — Research institute, Nancy, France · 32 works · 4,142 citations

📚 Shelves (5): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 18,614 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 17,965 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 17,450 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 17,296 works · [👁️ Computer Vision and Pattern Recognition](https://md.research.elamur.ai/free/en/shelves/computer-science/computer-vision-and-pattern-recognition/) · 3,486 works

🧭 Topics (1): [🌍 Multilingual & cross-lingual NLP](https://md.research.elamur.ai/free/en/topics/multilingual-nlp/) · 882 works

## Cite

Cite as: Yaya SY, Dioula Doucouré, Christophe Cerisara and Irina Illina. 2026. Cross-lingual Matryoshka Representation Learning across Speech and Text. Findings of the Association for Computational Linguistics: ACL 2026.

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{sy2026cross,
  title = {{Cross-lingual Matryoshka Representation Learning across Speech and Text}},
  author = {Yaya SY, Dioula Doucouré, Christophe Cerisara and Irina Illina},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.1710},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.1710}
}
```

- [← previous Cross-Lingual Empirical Evaluation of Large Language Models for Arabic Medical Tasks](https://md.research.elamur.ai/free/en/papers/abouzahir-2026-cross-lingual-empirical-evaluation-large/)
- [next → CrowS-Pairs-NL: A Benchmark to Evaluate Dutch Stereotype Bias in LLMs](https://md.research.elamur.ai/free/en/papers/weide-2026-crows-pairs-nl-benchmark-evaluate/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/11/)

## References

31 references, 6 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Luiz Henrique Bonifacio, Israel Campiotti, Roberto Lotufo, and Rodrigo Nogueira. 2021. mMARCO: A multilingual version of MS MARCO passage ranking dataset. arXiv:2108.13897. — [arXiv:2108.13897](https://arxiv.org/abs/2108.13897) [find](https://scholar.google.com/scholar?q=Luiz+Henrique+Bonifacio%2C+Israel+Campiotti%2C+Roberto+Lotufo%2C+and+Rodrigo+Nogueira%2E+2021%2E+mMARCO%3A+A+multilingual+version+of+MS+MARCO+passage+ranking+dataset%2E+arXiv%3A2108%2E13897%2E+%E2%80%94+%5BarXiv%3A2108%2E13897%5D%28https%3A)
2. Hervé Bredin. 2023. pyannote.audio 2.1 speaker diarization pipeline: principle, benchmark, and recipe. In Proc. [find](https://scholar.google.com/scholar?q=Herv%C3%A9+Bredin%2E+2023%2E+pyannote%2Eaudio+2%2E1+speaker+diarization+pipeline%3A+principle%2C+benchmark%2C+and+recipe%2E+In+Proc%2E)
3. Mu Cai, Jianwei Yang, Jianfeng Gao, and Yong Jae Lee. 2024. Matryoshka multimodal models. Preprint, arXiv:2405.17430. — [arXiv:2405.17430](https://arxiv.org/abs/2405.17430) [find](https://scholar.google.com/scholar?q=Mu+Cai%2C+Jianwei+Yang%2C+Jianfeng+Gao%2C+and+Yong+Jae+Lee%2E+2024%2E+Matryoshka+multimodal+models%2E+Preprint%2C+arXiv%3A2405%2E17430%2E+%E2%80%94+%5BarXiv%3A2405%2E17430%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F2405%2E17430%29)
4. Umberto Cappellazzo, Minsu Kim, and Stavros Petridis. 2025. Adaptive audio-visual speech recognition via matryoshka-based multimodal llms. Preprint, arXiv:2503.06362. — [arXiv:2503.06362](https://arxiv.org/abs/2503.06362) [find](https://scholar.google.com/scholar?q=Umberto+Cappellazzo%2C+Minsu+Kim%2C+and+Stavros+Petridis%2E+2025%2E+Adaptive+audio%2Dvisual+speech+recognition+via+matryoshka%2Dbased+multimodal+llms%2E+Preprint%2C+arXiv%3A2503%2E06362%2E+%E2%80%94+%5BarXiv%3A2503%2E06362%5D%28https%3A%2F%2Farxi)
5. Alexis Conneau, Kartikay Khandelwal, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer, and Veselin Stoyanov. 2019. [Unsupervised cross-lingual representation learning at scale](https://md.research.elamur.ai/free/en/papers/conneau-2020-unsupervised-cross-lingual-representation-learning/). CoRR, abs/1911.02116.
6. Abdou Karim Kandji, Frédéric Precioso, Cheikh Ba, Samba Ndiaye, and Augustin Ndione. 2025. Wolbanking77: Wolof banking speech intent classification dataset. Preprint, arXiv:2509.19271. — [arXiv:2509.19271](https://arxiv.org/abs/2509.19271) [find](https://scholar.google.com/scholar?q=Abdou+Karim+Kandji%2C+Fr%C3%A9d%C3%A9ric+Precioso%2C+Cheikh+Ba%2C+Samba+Ndiaye%2C+and+Augustin+Ndione%2E+2025%2E+Wolbanking77%3A+Wolof+banking+speech+intent+classification+dataset%2E+Preprint%2C+arXiv%3A2509%2E19271%2E+%E2%80%94+%5BarXiv%3A2509%2E1)
7. Aditya Kusupati, Gantavya Bhatt, Aniket Rege, Matthew Wallingford, Aditya Sinha, Vivek Ramanujan, William Howard-Snyder, Kaifeng Chen, Sham Kakade, Prateek Jain, and Ali Farhadi. 2022. Matryoshka representation learning. In Advances in Neural Information Processing Systems, volume 35, pages 30233–30… [find](https://scholar.google.com/scholar?q=Aditya+Kusupati%2C+Gantavya+Bhatt%2C+Aniket+Rege%2C+Matthew+Wallingford%2C+Aditya+Sinha%2C+Vivek+Ramanujan%2C+William+Howard%2DSnyder%2C+Kaifeng+Chen%2C+Sham+Kakade%2C+Prateek+Jain%2C+and+Ali+Farhadi%2E+2022%2E+Matryoshka+repr)
8. Tsz Kin Lam, Marco Gaido, Sara Papi, Luisa Bentivogli, and Barry Haddow. 2025. Prepending or cross-attention for speech-to-text? an empirical comparison. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Tech… [find](https://scholar.google.com/scholar?q=Tsz+Kin+Lam%2C+Marco+Gaido%2C+Sara+Papi%2C+Luisa+Bentivogli%2C+and+Barry+Haddow%2E+2025%2E+Prepending+or+cross%2Dattention+for+speech%2Dto%2Dtext%3F+an+empirical+comparison%2E+In+Proceedings+of+the+2025+Conference+of+the+N)
9. Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. 2023. [Visual instruction tuning](https://md.research.elamur.ai/free/en/papers/liu-2023-visual-instruction-tuning/). In Advances in Neural Information Processing Systems, volume 36, pages 34892–34916. Curran Associates, Inc.
10. Alexis Conneau, Min Ma, Simran Khanuja, Yu Zhang, Vera Axelrod, Siddharth Dalmia, Jason Riesa, Clara Rivera, and Ankur Bapna. 2022. [Fleurs: Few-shot learning evaluation of universal representations of speech](https://md.research.elamur.ai/free/en/papers/conneau-2022-fleurs-few-shot-learning-evaluation/). arXiv preprint arXiv:2205.12446.
11. Oriol Pareras, Gerard I. Gállego, Federico Costa, Cristina España-Bonet, and Javier Hernando. 2025. Revisiting direct speech-to-text translation with speech llms: Better scaling than cot prompting? Preprint, arXiv:2510.03093. — [arXiv:2510.03093](https://arxiv.org/abs/2510.03093) [find](https://scholar.google.com/scholar?q=Oriol+Pareras%2C+Gerard+I%2E+G%C3%A1llego%2C+Federico+Costa%2C+Cristina+Espa%C3%B1a%2DBonet%2C+and+Javier+Hernando%2E+2025%2E+Revisiting+direct+speech%2Dto%2Dtext+translation+with+speech+llms%3A+Better+scaling+than+cot+prompting%3F+Pr)
12. Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2018. [BERT: pre-training of deep bidirectional transformers for language understanding](https://md.research.elamur.ai/free/en/papers/devlin-2019-bert-pre-training-deep-bidirectional/). CoRR, abs/1810.04805.
13. Letian Peng, Yuwei Zhang, Zilong Wang, Jayanth Srinivasa, Gaowen Liu, Zihan Wang, and Jingbo Shang. 2024. Answer is all you need: Instruction-following text embedding via answering the question. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Lon… [find](https://scholar.google.com/scholar?q=Letian+Peng%2C+Yuwei+Zhang%2C+Zilong+Wang%2C+Jayanth+Srinivasa%2C+Gaowen+Liu%2C+Zihan+Wang%2C+and+Jingbo+Shang%2E+2024%2E+Answer+is+all+you+need%3A+Instruction%2Dfollowing+text+embedding+via+answering+the+question%2E+In+Pr)
14. Thierno Ibrahima DIOP. 2021. Wolof asr data on urban transport. [find](https://scholar.google.com/scholar?q=Thierno+Ibrahima+DIOP%2E+2021%2E+Wolof+asr+data+on+urban+transport%2E)
15. Benjamin Elizalde, Soham Deshmukh, Mahmoud Al Ismail, and Huaming Wang. 2022. Clap: Learning audio concepts from natural language supervision. Preprint, arXiv:2206.04769. [find](https://scholar.google.com/scholar?q=Benjamin+Elizalde%2C+Soham+Deshmukh%2C+Mahmoud+Al+Ismail%2C+and+Huaming+Wang%2E+2022%2E+Clap%3A+Learning+audio+concepts+from+natural+language+supervision%2E+Preprint%2C+arXiv%3A2206%2E04769%2E)
16. Elodie Gauthier, Laurent Besacier, Sylvie Voisin, Michael Melese, and Uriel Pascal Elingui. 2016. Collecting resources in sub-saharan african languages for automatic speech recognition: a case study of wolof. LREC. [find](https://scholar.google.com/scholar?q=Elodie+Gauthier%2C+Laurent+Besacier%2C+Sylvie+Voisin%2C+Michael+Melese%2C+and+Uriel+Pascal+Elingui%2E+2016%2E+Collecting+resources+in+sub%2Dsaharan+african+languages+for+automatic+speech+recognition%3A+a+case+study+o)
17. Elodie Gauthier, Aminata Ndiaye, and Abdoulaye Guissé. 2024. Kallaama: A transcribed speech dataset about agriculture in the three most widely spoken languages in Senegal. In Proceedings of the Fifth Workshop on Resources for African Indigenous Languages @ LREC-COLING 2024, pages 10–19, Torino, Ital… [find](https://scholar.google.com/scholar?q=Elodie+Gauthier%2C+Aminata+Ndiaye%2C+and+Abdoulaye+Guiss%C3%A9%2E+2024%2E+Kallaama%3A+A+transcribed+speech+dataset+about+agriculture+in+the+three+most+widely+spoken+languages+in+Senegal%2E+In+Proceedings+of+the+Fifth+)
18. Wenbo Hu, Zi-Yi Dou, Liunian Harold Li, Amita Kamath, Nanyun Peng, and Kai-Wei Chang. 2024. Matryoshka query transformer for large vision-language models. Preprint, arXiv:2405.19315. — [arXiv:2405.19315](https://arxiv.org/abs/2405.19315) [find](https://scholar.google.com/scholar?q=Wenbo+Hu%2C+Zi%2DYi+Dou%2C+Liunian+Harold+Li%2C+Amita+Kamath%2C+Nanyun+Peng%2C+and+Kai%2DWei+Chang%2E+2024%2E+Matryoshka+query+transformer+for+large+vision%2Dlanguage+models%2E+Preprint%2C+arXiv%3A2405%2E19315%2E+%E2%80%94+%5BarXiv%3A2405%2E193)
19. Alexis Plaquet and Hervé Bredin. 2023. Powerset multi-class cross entropy loss for neural speaker diarization. In Proc. [find](https://scholar.google.com/scholar?q=Alexis+Plaquet+and+Herv%C3%A9+Bredin%2E+2023%2E+Powerset+multi%2Dclass+cross+entropy+loss+for+neural+speaker+diarization%2E+In+Proc%2E)
20. Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. 2021. [Learning transferable visual models from natural language supervision](https://md.research.elamur.ai/free/en/papers/radford-2021-learning-transferable-visual-models-natural/). Preprint, arXiv:2103.00020.
21. Chandan K A Reddy, Vishak Gopal, and Ross Cutler. 2021. Dnsmos: A non-intrusive perceptual objective speech quality metric to evaluate noise suppressors. Preprint, arXiv:2010.15258. — [arXiv:2010.15258](https://arxiv.org/abs/2010.15258) [find](https://scholar.google.com/scholar?q=Chandan+K+A+Reddy%2C+Vishak+Gopal%2C+and+Ross+Cutler%2E+2021%2E+Dnsmos%3A+A+non%2Dintrusive+perceptual+objective+speech+quality+metric+to+evaluate+noise+suppressors%2E+Preprint%2C+arXiv%3A2010%2E15258%2E+%E2%80%94+%5BarXiv%3A2010%2E1525)
22. Fabian David Schmidt, Philipp Borchert, Ivan Vulić, and Goran Glavaš. 2024. Self-distillation for model stacking unlocks cross-lingual NLU in 200+ languages. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 6724–6743, Miami, Florida, USA. Association for Computational… [find](https://scholar.google.com/scholar?q=Fabian+David+Schmidt%2C+Philipp+Borchert%2C+Ivan+Vulic%CC%81%2C+and+Goran+Glava%C5%A1%2E+2024%2E+Self%2Ddistillation+for+model+stacking+unlocks+cross%2Dlingual+NLU+in+200%2B+languages%2E+In+Findings+of+the+Association+for+Comput)
23. Fabian David Schmidt, Ivan Vulić, Goran Glavaš, and David Ifeoluwa Adelani. 2025. Fleurs-slu: A massively multilingual benchmark for spoken language understanding. Preprint, arXiv:2501.06117. — [arXiv:2501.06117](https://arxiv.org/abs/2501.06117) [find](https://scholar.google.com/scholar?q=Fabian+David+Schmidt%2C+Ivan+Vulic%CC%81%2C+Goran+Glava%C5%A1%2C+and+David+Ifeoluwa+Adelani%2E+2025%2E+Fleurs%2Dslu%3A+A+massively+multilingual+benchmark+for+spoken+language+understanding%2E+Preprint%2C+arXiv%3A2501%2E06117%2E+%E2%80%94+%5BarXi)
24. Yaya Sy, Dioula Doucouré, Christophe Cerisara, and Irina Illina. 2025. Speech language models for under-represented languages: Insights from wolof. Preprint, arXiv:2509.15362. — [arXiv:2509.15362](https://arxiv.org/abs/2509.15362) [find](https://scholar.google.com/scholar?q=Yaya+Sy%2C+Dioula+Doucour%C3%A9%2C+Christophe+Cerisara%2C+and+Irina+Illina%2E+2025%2E+Speech+language+models+for+under%2Drepresented+languages%3A+Insights+from+wolof%2E+Preprint%2C+arXiv%3A2509%2E15362%2E+%E2%80%94+%5BarXiv%3A2509%2E15362%5D%28htt)
25. Yuqing Tang, Chau Tran, Xian Li, Peng-Jen Chen, Naman Goyal, Vishrav Chaudhary, Jiatao Gu, and Angela Fan. 2020. Multilingual translation with extensible multilingual pretraining and finetuning. [find](https://scholar.google.com/scholar?q=Yuqing+Tang%2C+Chau+Tran%2C+Xian+Li%2C+Peng%2DJen+Chen%2C+Naman+Goyal%2C+Vishrav+Chaudhary%2C+Jiatao+Gu%2C+and+Angela+Fan%2E+2020%2E+Multilingual+translation+with+extensible+multilingual+pretraining+and+finetuning%2E)
26. Silero Team. 2024. Silero vad: pre-trained enterprise-grade voice activity detector (vad), number detector and language classifier. [https://github.com/](https://github.com/) snakers4/silero-vad. [↗](https://github.com/)
27. Lewis Tunstall, Nils Reimers, Unso Eun Seo Jo, Luke Bates, Daniel Korat, Moshe Wasserblat, and Oren Pereg. 2022. Efficient few-shot learning without prompts. arXiv preprint. [find](https://scholar.google.com/scholar?q=Lewis+Tunstall%2C+Nils+Reimers%2C+Unso+Eun+Seo+Jo%2C+Luke+Bates%2C+Daniel+Korat%2C+Moshe+Wasserblat%2C+and+Oren+Pereg%2E+2022%2E+Efficient+few%2Dshot+learning+without+prompts%2E+arXiv+preprint%2E)
28. Aaron van den Oord, Yazhe Li, and Oriol Vinyals. 2019. [Representation learning with contrastive predictive coding](https://md.research.elamur.ai/free/en/papers/oord-2018-representation-learning-contrastive-predictive-coding/). Preprint, arXiv:1807.03748.
29. Tiansheng Wen, Yifei Wang, Zequn Zeng, Zhong Peng, Yudi Su, Xinyang Liu, Bo Chen, Hongwei Liu, Stefanie Jegelka, and Chenyu You. 2025. Beyond matryoshka: Revisiting sparse coding for adaptive representation. Preprint, arXiv:2503.01776. — [arXiv:2503.01776](https://arxiv.org/abs/2503.01776) [find](https://scholar.google.com/scholar?q=Tiansheng+Wen%2C+Yifei+Wang%2C+Zequn+Zeng%2C+Zhong+Peng%2C+Yudi+Su%2C+Xinyang+Liu%2C+Bo+Chen%2C+Hongwei+Liu%2C+Stefanie+Jegelka%2C+and+Chenyu+You%2E+2025%2E+Beyond+matryoshka%3A+Revisiting+sparse+coding+for+adaptive+represen)
30. Jian Wu, Yashesh Gaur, Zhuo Chen, Long Zhou, Yimeng Zhu, Tianrui Wang, Jinyu Li, Shujie Liu, Bo Ren, Linquan Liu, and Yu Wu. 2023. On decoder-only architecture for speech-to-text and large language model integration. Preprint, arXiv:2307.03917. — [arXiv:2307.03917](https://arxiv.org/abs/2307.03917) [find](https://scholar.google.com/scholar?q=Jian+Wu%2C+Yashesh+Gaur%2C+Zhuo+Chen%2C+Long+Zhou%2C+Yimeng+Zhu%2C+Tianrui+Wang%2C+Jinyu+Li%2C+Shujie+Liu%2C+Bo+Ren%2C+Linquan+Liu%2C+and+Yu+Wu%2E+2023%2E+On+decoder%2Donly+architecture+for+speech%2Dto%2Dtext+and+large+language+mo)
31. Biao Zhang, Lixin Chen, Tong Liu, and Bo Zheng. 2025. SMEC:rethinking matryoshka representation learning for retrieval embedding compression. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, pages 26220–26233, Suzhou, China. Association for Computational Lin… [find](https://scholar.google.com/scholar?q=Biao+Zhang%2C+Lixin+Chen%2C+Tong+Liu%2C+and+Bo+Zheng%2E+2025%2E+SMEC%3Arethinking+matryoshka+representation+learning+for+retrieval+embedding+compression%2E+In+Proceedings+of+the+2025+Conference+on+Empirical+Methods)

*Updated 2 November 2026*