Michał Tyrolski

Michał Tyrolski

Senior AI Consultant at EY / Independent Researcher.

Posts by Collection

other_papers

Explainable Machine Learning at Microsoft MLADS

Published:

Had the privilege to lead explainable AI research at Microsoft Ireland during an internship, utilizing petabytes of data to develop interpretable machine learning solutions for both theoretical and practical applications, presented at the internal MLADS conference.

Recommended citation: Tyrolski, M. (2022). Explainable machine learning at Microsoft MLADS (Internal Microsoft MLADS Conference presentation).
Download Paper | Download Slides

Enhancing Antigenic Peptide Discovery

Published:

This study pinpoints evaluation pitfalls in MHC-I presentation prediction and proposes a unified framework to standardize methodology. It also introduces a transformer model trained on interspecies data, markedly improving peptide–MHC-I binding accuracy and generalization across diverse peptides, alleles, and proteins.

Recommended citation: Giziński, S., Preibisch, G., Kucharski, P., Tyrolski, M., Rembalski, M., Grzegorczyk, P., & Gambin, A. (2024). Enhancing antigenic peptide discovery: Improved MHC-I binding prediction and methodology. Methods, 224, 1–9. https://doi.org/10.1016/j.ymeth.2024.01.016
Download Paper | Download Slides

publications

Hierarchical Transformers Are More Efficient Language Models

Published in NAACL 2022, 2022

Hourglass: a hierarchical Transformer with down/upsampling layers that improves long-sequence modeling efficiency; SOTA on ImageNet32 and strong enwik8 performance.

Recommended citation: Nawrot, P., Tworkowski, S., Tyrolski, M., Kaiser, Ł., Wu, Y., Szegedy, C., & Michalewski, H. (2021). Hierarchical transformers are more efficient language models. arXiv preprint arXiv:2110.13711.
Download Paper | Download Slides

Adjusting Planning Horizon with Adaptive Subgoal Search

Published in ICLR 2023 (Top-5%, Oral), 2023

AdaSubS adaptively adjusts planning horizon via diverse subgoals + fast reachability filtering; efficient on Sokoban, Rubik’s Cube, INT.

Recommended citation: Zawalski, M., Tyrolski, M., Czechowski, K., Odrzygóźdź, T., Stachura, D., Piękos, P., Wu, Y., Kuciński, Ł., & Miłoś, P. (2022). Fast and precise: Adjusting planning horizon with adaptive subgoal search. arXiv preprint arXiv:2206.00702.
Download Paper | Download Slides

What Matters in Hierarchical Search for Combinatorial Problems?

Published in ICLR 2024 (Generative Models for Decision Making), 2024

Empirical analysis of properties shaping hierarchical search in combinatorial reasoning; guidelines for robust comparison + future design.

Recommended citation: Zawalski, M., Góral, G., Tyrolski, M., Wiśnios, E., Budrowski, F., Cygan, M., Kuciński, Ł., & Miłoś, P. (2024). What matters in hierarchical search for combinatorial reasoning problems? arXiv preprint arXiv:2406.03361.
Download Paper | Download Slides

OpenGVL: Benchmarking Visual Temporal Progress for Data Curation

Published in CoRL 2025 workshop, 2025

Benchmark and toolkit for evaluating VLMs’ sense of progress in robotics via Value-Order Correlation (VOC); enables automated dataset curation from videos.

Recommended citation: Budzianowski, P., Wiśnios, E., Góral, G., Tyrolski, M., Kulakov, I., Petrenko, V., & Walas, K. (2025). OpenGVL - Benchmarking Visual Temporal Progress for Data Curation. arXiv:2509.17321.
Download Paper | Download Slides

talks