Dmitry Baranchuk

prof_pic.jpg

I am a Lead Research Scientist at Yandex Research. My work broadly focuses on diffusion-based models and exploring effective synergies between different generative paradigms. One of my key directions is making large-scale visual generative models more efficient and practical for real-world use. I am also highly interested in extending the use of generative models to various downstream tasks, e.g., leveraging them as strong feature extractors or data engines.

In addition to my research, I organize seminars on generative models, post paper reviews, and teach the Visual GenAI course at the Yandex School of Data Analysis.

Previously: I received my Ph.D. in Computer Science at HSE University in December 2024, advised by Artem Babenko. In 2022, I was a research intern at Meta AI, where I worked on large-scale vector search and retrieval-augmented LLMs, advised by Matthijs Douze and Zeki Yalniz. Earlier, in 2019, I interned at ETH Zurich, working with Vincent Fortuin and Stephan Mandt on generative modeling for missing data imputation.

Selected Papers

Generative Modeling
  1. ICLR
    Rethinking Global Text Conditioning in Diffusion Transformers
    Nikita Starodubcev, Daniil Pakhomov, Zongze Wu, Ilya Drobyshevskiy, Yuchen Liu, Zhonghao Wang, Yuqian Zhou, Zhe Lin, and Dmitry Baranchuk
    Feb 2026
    HTML PDF
  2. ICLR
    Scale-wise Distillation of Diffusion Models
    Nikita Starodubcev, Ilya Drobyshevskiy, Denis Kuznedelev, Artem Babenko, and Dmitry Baranchuk
    Feb 2026
    HTML PDF
  3. CVPR
    Switti: Designing Scale-Wise Transformers for Text-to-Image Synthesis
    Anton Voronov, Denis Kuznedelev, Mikhail Khoroshikh, Valentin Khrulkov, and Dmitry Baranchuk
    In Computer Vision and Pattern Recognition (CVPR), Jun 2025
    HTML PDF
  4. NeurIPS
    Invertible Consistency Distillation for Text-Guided Image Editing in Around 7 Steps
    Nikita Starodubcev, Mikhail Khoroshikh, Artem Babenko, and Dmitry Baranchuk
    In Neural Information Processing Systems, Dec 2024
    HTML PDF
  5. arXiv
    Accurate Compression of Text-to-Image Diffusion Models via Vector Quantization
    Vage Egiazarian*, Denis Kuznedelev*, Anton Voronov*, Ruslan Svirschevski, Michael Goin, Daniil Pavlov, Dan Alistarh, and Dmitry Baranchuk
    Sep 2024
    HTML PDF
  6. CVPR
    Your Student is Better Than Expected: Adaptive Teacher-Student Collaboration for Text-Conditional Diffusion Models
    Nikita Starodubcev, Artem Fedorov, Artem Babenko, and Dmitry Baranchuk
    In Computer Vision and Pattern Recognition (CVPR), Jun 2024
    HTML PDF
  7. ICML
    TabDDPM: Modelling Tabular Data with Diffusion Models
    Akim Kotelnikov, Dmitry Baranchuk, Ivan Rubachev, and Artem Babenko
    In International Conference on Machine Learning, Jul 2023
    HTML PDF
  8. ACL Demo
    Petals: Collaborative Inference and Fine-tuning of Large Models
    Alexander Borzunov*, Dmitry Baranchuk*, Tim Dettmers*, Maksim Riabinin*, Younes Belkada*, Artem Chumachenko, Pavel Samygin, and Colin Raffel
    In Association for Computational Linguistics (System Demonstrations), Jul 2023
    HTML PDF
  9. ICLR
    Label-Efficient Semantic Segmentation with Diffusion Models
    Dmitry Baranchuk, Andrey Voynov, Ivan Rubachev, Valentin Khrulkov, and Artem Babenko
    In International Conference on Learning Representations, May 2022
    HTML PDF
  10. AISTATS
    GP-VAE: Deep Probabilistic Time Series Imputation
    Vincent Fortuin*, Dmitry Baranchuk*, Gunnar Raetsch, and Stephan Mandt
    In International Conference on Artificial Intelligence and Statistics, Aug 2020
    HTML PDF
Large-scale Vector Search
  1. ICCV
    DeDrift: Robust Similarity Search under Content Drift
    Dmitry Baranchuk, Matthijs Douze, Yash Upadhyay, and I. Zeki Yalniz
    In International Conference on Computer Vision (ICCV), Oct 2023
    HTML PDF
  2. NeurIPS Competition
    Results of the NeurIPS’21 Challenge on Billion-Scale Approximate Nearest Neighbor Search
    Harsha Vardhan Simhadri, George Williams, Martin Aumüller, Matthijs Douze, Artem Babenko, Dmitry Baranchuk, Qi Chen, Lucas Hosseini, Ravishankar Krishnaswamny, Gopal Srinivasa, Suhas Jayaram Subramanya, and Jingdong Wang
    In NeurIPS 2021 Competitions and Demonstrations Track, Dec 2022
    HTML PDF
  3. ICML
    Learning to Route in Similarity Graphs
    Dmitry Baranchuk, Dmitry Persiyanov, Anton Sinitsin, and Artem Babenko
    In International Conference on Machine Learning, Jun 2019
    HTML PDF
  4. ECCV
    Revisiting the Inverted Indices for Billion-Scale Approximate Nearest Neighbors
    Dmitry Baranchuk, Artem Babenko, and Yury Malkov
    In European Conference on Computer Vision (ECCV), Sep 2018
    HTML PDF