Michal Byra

Samsung AI Center, Polish Academy of Sciences

Papers on Lattice

Total citations

Topics

Publication activitypapers/week, last 8 weeks

Research focus

Architecture Design (Transformers, SSMs, MoE) (1)Computer Vision (1)Training Efficiency & Optimization (1)

Frequent co-authors

Grzegorz Gruszczynski (1)Pawel Olszowiec (1)Grzegorz Stefanski (1)Alberto Presta (1)

Papers (1)

Aug 5, 2026

3w ago·also Polish Academy of Sciences

Training Crossroads for Recurrent Vision Transformers: Recurrence, Neural ODEs, and Deep Supervision

Recurrent Vision Transformers can outperform standard models in accuracy-to-parameter trade-offs when memory constraints are prioritized, challenging conventional wisdom about architectural efficiency.

Grzegorz Gruszczynski, Pawel Olszowiec, Michal Byra +2

Architecture Design (Transformers, SSMs, MoE)Computer Vision Training Efficiency & Optimization

Search

Michal Byra

Publication activitypapers/week, last 8 weeks

Research focus

Frequent co-authors

Papers (1)