---
title: "mllm-shap — paper by Muszyński et al. (2026)"
description: "The authors present mllm-shap, an open-source, MIT-licensed Python package with a web GUI that fills this… 90% accuracy. 20 references."
authors: ["Jakub Muszyński", "Paweł Pozorski", "Maria Ganzha"]
year: 2026
venue: "Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 3: System Demonstrations)"
source: "https://aclanthology.org/2026.acl-demo.38/"
doi: "10.18653/v1/2026.acl-demo.38"
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/muszynski-2026-mllm-shap-shapley-value-explainability/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/muszynski-2026-mllm-shap-shapley-value-explainability/)

[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/42/) / mllm-shap: A Shapley Value Explainability Platform for Text-Audio Multimodal Large Language Models

# mllm-shap: A Shapley Value Explainability Platform for Text-Audio Multimodal Large Language Models

**Authors:** Jakub Muszyński, Paweł Pozorski and Maria Ganzha · **Year:** 2026 · **Venue:** [Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 3: System Demonstrations)](https://md.research.elamur.ai/free/en/venues/acl/) ([ACL 2026](https://md.research.elamur.ai/free/en/venues/acl/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.acl-demo.38/) · [PDF](https://aclanthology.org/2026.acl-demo.38.pdf) · [DOI 10.18653/v1/2026.acl-demo.38](https://doi.org/10.18653/v1/2026.acl-demo.38) · [OpenAlex](https://openalex.org/W7166818731)

**Citations:** 0 citations (OpenAlex, 2026-10-14)

## Summary

Shapley-value attribution tools such as SHAP and TokenSHAP handle tabular, image or text-only inputs. They cannot explain multimodal LLMs that take speech and text together. The authors present mllm-shap, an open-source, MIT-licensed Python package with a web GUI that fills this gap for text-audio MLLMs. It needs only black-box access through a generate API.

The package addresses three problems:
- **Mixed inputs:** it treats text tokens and audio frames differently within one coalition game. Text tokens are dropped, while audio is silenced by zeroing the waveform or frames inside segment boundaries.
- **Conversation structure:** every token carries metadata for modality, role and turn, so multi-turn dialogues can be analysed and system tokens left out.
- **Too many audio units:** dense audio frames are merged into word-level units with SGPA phonetic alignment, which shrinks the coalition space.

Five SV estimators are offered. A Neyman-allocated variant of Complementary Contributions gives the best accuracy. Demonstrations on LFM2-Audio-1.5B show sensible attributions; in a transcription task, for example, the audio input receives most of the credit. Overall, the tool makes principled multimodal attribution practical on consumer hardware.

## Key points

- **Estimators:** Exact, Monte Carlo, Complementary Contributions, Neyman-CC and Hierarchical. On 27 inputs of 9–10 tokens where exact SVs are computable, Neyman-CC exceeded 90% accuracy at a sampling fraction of only 0.15.
- **SGPA grouping:** it uses CTC alignment with Wav2Vec2-XLSR-53 plus spectrogram-based boundary refinement. It cut about 50 audio tokens to about 7 segments, needing 43× fewer model calls and reducing runtime from about 30 minutes to about 1 minute per sample.
- **Evaluation scale:** 593 SV analyses were run on a single consumer GPU, using the VoiceBench and Infinity-Instruct datasets.
- **Interface and distribution:** the GUI uses a React frontend and a FastAPI backend, with cost estimation and session persistence. The package is installable from PyPI, and the code is on GitHub.
- **Limitations:** only one model, LFM2-Audio-1.5B, was tested. Long inputs still take about 400 s each. There is no faithfulness comparison against LIME or Integrated Gradients.

*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

- [Warsaw University of Technology](https://md.research.elamur.ai/free/en/universities/warsaw-university-of-technology/) — University, Warsaw, Poland · 3 works · 256 citations

📚 Shelves (3): [🖥️ 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 · [👁️ Computer Vision and Pattern Recognition](https://md.research.elamur.ai/free/en/shelves/computer-science/computer-vision-and-pattern-recognition/) · 3,486 works

🧭 Topics (4): [🧠 Language models & pretraining](https://md.research.elamur.ai/free/en/topics/language-models/) · 3,894 works · [🖼️ Vision & language](https://md.research.elamur.ai/free/en/topics/multimodal/) · 811 works · [🎙️ Speech & spoken language](https://md.research.elamur.ai/free/en/topics/speech/) · 519 works · [🔬 Interpretability](https://md.research.elamur.ai/free/en/topics/interpretability/) · 149 works

## Cite

Cite as: Jakub Muszyński, Paweł Pozorski and Maria Ganzha. 2026. mllm-shap: A Shapley Value Explainability Platform for Text-Audio Multimodal Large Language Models. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 3: System Demonstrations).

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

```bibtex
@inproceedings{muszynski2026mllm,
  title = {{mllm-shap: A Shapley Value Explainability Platform for Text-Audio Multimodal Large Language Models}},
  author = {Jakub Muszyński, Paweł Pozorski and Maria Ganzha},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 3: System Demonstrations)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-demo.38},
  url = {https://doi.org/10.18653/v1/2026.acl-demo.38}
}
```

- [← previous MLingualFC: Evaluating Jailbreak Vulnerabilities in Multilingual Vision-Language Models](https://md.research.elamur.ai/free/en/papers/makwana-2026-mlingualfc-evaluating-jailbreak-vulnerabilities-multilingual/)
- [next → MM-ShiftKV: Decode-Aware Prefill-Stage KV Selection for Multimodal Large Language Models](https://md.research.elamur.ai/free/en/papers/shu-2026-mm-shiftkv-decode-aware-prefill/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/42/)

## References

20 references, 7 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Alexei Baevski, Yuhao Zhou, Abdelrahman Mohamed, and Michael Auli. 2020. [wav2vec 2.0: A framework for self-supervised learning of speech representations](https://md.research.elamur.ai/free/en/papers/baevski-2020-wav2vec-2-0-framework-self/). In Advances in Neural Information Processing Systems, volume 33, pages 12449–12460.
2. Yiming Chen, Xianghu Yue, Chen Zhang, Xiaoxue Gao, Robby T. Tan, and Haizhou Li. 2024. [Voicebench: Benchmarking LLM-based voice assistants](https://md.research.elamur.ai/free/en/papers/chen-2026-voicebench-benchmarking-llm-based-voice/). arXiv preprint arXiv:2410.17196.
3. Ian Covert, Scott M Lundberg, and Su-In Lee. 2021. Explaining by removing: A unified framework for model explanation. Journal of Machine Learning Research, 22(209):1–90. [find](https://scholar.google.com/scholar?q=Ian+Covert%2C+Scott+M+Lundberg%2C+and+Su%2DIn+Lee%2E+2021%2E+Explaining+by+removing%3A+A+unified+framework+for+model+explanation%2E+Journal+of+Machine+Learning+Research%2C+22%28209%29%3A1%E2%80%9390%2E)
4. Roni Goldshmidt and Miriam Horovicz. 2024. To-kenshap: Interpreting large language models with monte carlo shapley value estimation. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing. [find](https://scholar.google.com/scholar?q=Roni+Goldshmidt+and+Miriam+Horovicz%2E+2024%2E+To%2Dkenshap%3A+Interpreting+large+language+models+with+monte+carlo+shapley+value+estimation%2E+In+Proceedings+of+the+2024+Conference+on+Empirical+Methods+in+Natur)
5. Alex Graves, Santiago Fernández, Faustino Gomez, and Jürgen Schmidhuber. 2006. Connectionist temporal classification: Labelling unsegmented sequence data with recurrent neural networks. In Proceedings of the 23rd International Conference on Machine Learning, pages 369–376. [find](https://scholar.google.com/scholar?q=Alex+Graves%2C+Santiago+Ferna%CC%81ndez%2C+Faustino+Gomez%2C+and+Ju%CC%88rgen+Schmidhuber%2E+2006%2E+Connectionist+temporal+classification%3A+Labelling+unsegmented+sequence+data+with+recurrent+neural+networks%2E+In+Proceedin)
6. Sarthak Jain and Byron C. Wallace. 2019. [Attention is not explanation](https://md.research.elamur.ai/free/en/papers/jain-2019-attention-is-not-explanation/). In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 3543–3556.
7. Jijie Li, Li Du, Hanyu Zhao, Bo-wen Zhang, Liangdong Wang, Boyan Gao, Guang Liu, and Yonghua Lin. 2025. Infinity instruct: Scaling instruction selection and synthesis to enhance language models. arXiv preprint arXiv:2506.11116. — [arXiv:2506.11116](https://arxiv.org/abs/2506.11116) [find](https://scholar.google.com/scholar?q=Jijie+Li%2C+Li+Du%2C+Hanyu+Zhao%2C+Bo%2Dwen+Zhang%2C+Liangdong+Wang%2C+Boyan+Gao%2C+Guang+Liu%2C+and+Yonghua+Lin%2E+2025%2E+Infinity+instruct%3A+Scaling+instruction+selection+and+synthesis+to+enhance+language+models%2E+arXiv)
8. Liquid AI. 2025. Lfm2-audio: Liquid foundation models for audio. [https://www.liquid.ai/liquid-foundation-models/](https://www.liquid.ai/liquid-foundation-models/) lfm-2-audio. [↗](https://www.liquid.ai/liquid-foundation-models/)
9. Scott M Lundberg and Su-In Lee. 2017. [A unified approach to interpreting model predictions](https://md.research.elamur.ai/free/en/papers/lundberg-2017-unified-approach-interpreting-model-predictions/). In Advances in Neural Information Processing Systems, volume 30.
10. Meta Platforms. 2013. React: A JavaScript library for building user interfaces. [https://github.com/facebook/react](https://github.com/facebook/react). [↗](https://github.com/facebook/react)
11. Rory Mitchell, Joshua Cooper, Eibe Frank, and Geoffrey Holmes. 2022. Approximating the shapley value without marginal contributions. Proceedings of the AAAI Conference on Artificial Intelligence, 36(7):7773–7780. [find](https://scholar.google.com/scholar?q=Rory+Mitchell%2C+Joshua+Cooper%2C+Eibe+Frank%2C+and+Geoffrey+Holmes%2E+2022%2E+Approximating+the+shapley+value+without+marginal+contributions%2E+Proceedings+of+the+AAAI+Conference+on+Artificial+Intelligence%2C+36%287)
12. Edoardo Mosca, Ferenc Szigeti, Stella Tragianni, Daniel Gallagher, and Georg Groh. 2022. SHAP-based explanation methods: A review for NLP interpretability. In Proceedings of the 29th International Conference on Computational Linguistics, pages 4593–4603. [find](https://scholar.google.com/scholar?q=Edoardo+Mosca%2C+Ferenc+Szigeti%2C+Stella+Tragianni%2C+Daniel+Gallagher%2C+and+Georg+Groh%2E+2022%2E+SHAP%2Dbased+explanation+methods%3A+A+review+for+NLP+interpretability%2E+In+Proceedings+of+the+29th+International+Con)
13. Jerzy Neyman. 1934. On the two different aspects of the representative method: The method of stratified sampling and the method of purposive selection. Journal of the Royal Statistical Society, 97(4):558– 625. [find](https://scholar.google.com/scholar?q=Jerzy+Neyman%2E+1934%2E+On+the+two+different+aspects+of+the+representative+method%3A+The+method+of+stratified+sampling+and+the+method+of+purposive+selection%2E+Journal+of+the+Royal+Statistical+Society%2C+97%284%29%3A)
14. Paweł Pozorski, Jakub Muszyński, and Maria Ganzha. 2026. Sgpa: Spectrogram-guided phonetic alignment for feasible shapley value explanations in multimodal large language models. Preprint, arXiv:2603.02250. Submitted to Interspeech 2026. fixed budget. The estimator operates in two phases: an initial… — [arXiv:2603.02250](https://arxiv.org/abs/2603.02250) [find](https://scholar.google.com/scholar?q=Pawe%C5%82+Pozorski%2C+Jakub+Muszyn%CC%81ski%2C+and+Maria+Ganzha%2E+2026%2E+Sgpa%3A+Spectrogram%2Dguided+phonetic+alignment+for+feasible+shapley+value+explanations+in+multimodal+large+language+models%2E+Preprint%2C+arXiv%3A2603%2E)
15. Nils Reimers and Iryna Gurevych. 2019. [Sentence-BERT: Sentence embeddings using siamese BERT-networks](https://md.research.elamur.ai/free/en/papers/reimers-2019-sentence-bert-sentence-embeddings-siamese/). In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing, pages 3973–3983.
16. Hierarchical: A recursive decomposition strategy designed for long sequences that exceed the computational limits of flat estimators. Tokens are grouped into higher-level features, SVs are computed at the group level, and then recursively refined within groups of interest. [find](https://scholar.google.com/scholar?q=Hierarchical%3A+A+recursive+decomposition+strategy+designed+for+long+sequences+that+exceed+the+computational+limits+of+flat+estimators%2E+Tokens+are+grouped+into+higher%2Dlevel+features%2C+SVs+are+computed+at)
17. Marco Tulio Ribeiro, Sameer Singh, and Carlos Guestrin. 2016. [“why should I trust you?”: Explaining the predictions of any classifier](https://md.research.elamur.ai/free/en/papers/ribeiro-2016-why-should-i-trust-explaining/). In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 1135–1144.
18. Lloyd S. Shapley. 1951. Notes on the N-Person Game; II: The Value of an N-Person Game. RAND Corporation, Santa Monica, CA. [find](https://scholar.google.com/scholar?q=Lloyd+S%2E+Shapley%2E+1951%2E+Notes+on+the+N%2DPerson+Game%3B+II%3A+The+Value+of+an+N%2DPerson+Game%2E+RAND+Corporation%2C+Santa+Monica%2C+CA%2E)
19. Mukund Sundararajan, Ankur Taly, and Qiqi Yan. 2017. Axiomatic attribution for deep networks. In Proceedings of the 34th International Conference on Machine Learning, pages 3319–3328. [find](https://scholar.google.com/scholar?q=Mukund+Sundararajan%2C+Ankur+Taly%2C+and+Qiqi+Yan%2E+2017%2E+Axiomatic+attribution+for+deep+networks%2E+In+Proceedings+of+the+34th+International+Conference+on+Machine+Learning%2C+pages+3319%E2%80%933328%2E)
20. Haiyan Zhao, Hanjie Chen, Fan Yang, Ninghao Liu, Huiqi Deng, Hengyi Cai, Shuaiqiang Wang, Dawei Yin, and Mengnan Du. 2024. [Explainability for large language models: A survey](https://md.research.elamur.ai/free/en/papers/zhao-2023-explainability-large-language-models-survey/). ACM Transactions on Intelligent Systems and Technology, 15(2).

*Updated 1 November 2026*