Bartosz Wójcik

PhD student, Jagiellonian University

Joined

May 2021

Names

Bartosz Wójcik (Preferred)

Emails

****@gmail.com (Confirmed)

****@ideas-ncbr.pl (Confirmed)

****@doctoral.uj.edu.pl (Confirmed)

Personal Links

Google Scholar

ORCID

Career & Education History

PhD student

Jagiellonian University (uj.edu.pl)

2019 – Present

PhD student

IDEAS NCBR (ideas-ncbr.pl)

2022 – 2025

Researcher

Cracow University of Technology (pk.edu.pl)

2017 – 2019

Advisors, Relations & Conflicts

PhD Advisor

Marek Śmieja

2023 – Present

PhD Advisor

Jacek Tabor

2019 – 2023

Expertise

Uncertainty

2020 – Present

Conditional computation

2020 – Present

Deep learning

2017 – Present

Adversarial examples

2019 – 2022

Publications

Efficient Multi-Source Knowledge Transfer by Model Merging
Marcin Osial, Bartosz Wójcik, Bartosz Michał Zieliński, Sebastian Cygert
- Submitted to ICLR 2026
- Readers: Everyone
Efficient Large Language Models Moderation with Multi-Layer Latent Prototypes
Maciej Chrabaszcz, Filip Szatkowski, Bartosz Wójcik, Jan Dubiński, Tomasz Trzcinski, Sebastian Cygert
- ICLR 2026 Conference Desk Rejected Submission
- Readers: Everyone
Universal Properties of Activation Sparsity in Modern Large Language Models
Filip Szatkowski, Patryk Będkowski, Alessio Devoto, Jan Dubiński, Pasquale Minervini, Mikołaj Piórczyński, Simone Scardapane, Bartosz Wójcik
- ICLR 2026 Poster
- Readers: Everyone
Failure Prediction Is a Better Performance Proxy for Early-Exit Networks Than Calibration
Piotr Kubaty, Filip Szatkowski, Metod Jazbec, Bartosz Wójcik
- NeurIPS 2025 - Reliable ML Workshop
- Readers: Everyone
Failure Prediction Is a Better Performance Proxy for Early-Exit Networks Than Calibration
Piotr Kubaty, Filip Szatkowski, Metod Jazbec, Bartosz Wójcik
- SPIGM @ NeurIPS
- Readers: Everyone
Universal Properties of Activation Sparsity in Modern Large Language Models
Filip Szatkowski, Patryk Będkowski, Alessio Devoto, Jan Dubiński, Pasquale Minervini, Mikołaj Piórczyński, Simone Scardapane, Bartosz Wójcik
- NeurIPS 2025 ER Workshop
- Readers: Everyone
Universal Properties of Activation Sparsity in Modern Large Language Models
Filip Szatkowski, Patryk Będkowski, Alessio Devoto, Jan Dubiński, Pasquale Minervini, Mikołaj Piórczyński, Simone Scardapane, Bartosz Wójcik
- UniReps2025
- Readers: Everyone
Maybe I Should Not Answer That, but... Do LLMs Understand The Safety of Their Inputs?
Maciej Chrabaszcz, Filip Szatkowski, Bartosz Wójcik, Jan Dubiński, Tomasz Trzcinski
- BuildingTrust
- Readers: Everyone
How to Train Your Multi-Exit Model? Analyzing the Impact of Training Strategies
Piotr Kubaty, Bartosz Wójcik, Bartłomiej Tomasz Krzepkowski, Monika Michaluk, Tomasz Trzcinski, Jary Pomponi, Kamil Adamczewski
- ICML 2025 poster
- Readers: Everyone
Joint or Disjoint: Mixing Training Regimes for Early-Exit Models
Piotr Kubaty, Bartłomiej Tomasz Krzepkowski, Bartosz Wójcik, Monika Michaluk, Franciszek Szarwacki, Tomasz Trzcinski, Jary Pomponi, Kamil Adamczewski
- Submitted to ICLR 2025
- Readers: Everyone

View all 14 publications