Author

Bartłomiej Twardowski

h-index 18 View on Semantic Scholar

We have 4 of 58 papers

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

#machine learning Preprint Oct 2026

On the Limits of Representation-Based Test-Time Adaptation for Multimodal LLMs

Vision-language models suffer significant accuracy drops under common corruptions. While test-time adaptation (TTA) successfully mitigates these shifts in CLIP-like contrastive models, its application to generative multimodal large language models (MLLMs) remains underexplored and uniquely challenging due to their sequ...

Mikolaj Piórczynski, Wojciech Masarczyk, Damian Sójka et al. · 0 citations

#machine learning Preprint Oct 2026

Does Physics Live in the Activations? Localizing Physical Quantities in Video Diffusion Models

Video generation models produce strikingly realistic sequences and are increasingly proposed as world models, yet recent benchmarks reveal pronounced deficits in their physical reasoning. This raises the question of whether these models internalize physical principles or merely reproduce familiar motion patterns. We ad...

Jonas Kneifl, J. Skalski, Bartłomiej Twardowski et al. · 0 citations

#machine learning Preprint Sep 2026

Rethinking the Teacher-Student Framework for Test-Time Adaptation

This work challenges the common strategy of setting the teacher weights to an exponential moving average of the student by showing that error accumulation still occurs, although it is mostly apparent on longer sequences compared to those commonly utilized, and proposes an intransigent teacher that does not update its w...

Damian Sójka, Marc Masana, Bartłomiej Twardowski et al. · 0 citations

Review Jul 2026

Lifelong Representations: A Survey on Continual Self-Supervised Learning for Vision Models

This work presents a comprehensive survey of CSSL for vision, with connections to emerging vision-language settings and argues that advancing CSSL requires moving beyond small-scale benchmarks towards continual pre-training paradigms for large-scale systems.

Sergi Masip Cabeza, Alicja Dobrzeniecka, Jonathan Swinnen et al. · 0 citations