Tomasz Trzcinski

Efficient Large Language Models Moderation with Multi-Layer Latent Prototypes

Unpacking Softmax: How Logits Norm Drives Representation Collapse, Compression and Generalization

1000 Layer Networks for Self-Supervised RL: Scaling Depth Can Enable New Goal-Reaching Capabilities

Improving Continual Learning Performance and Efficiency with Auxiliary Classifiers

How to Train Your Multi-Exit Model? Analyzing the Impact of Training Strategies

Test-time Contrastive Concepts for Open-World Semantic Segmentation

Auxiliary Classifiers Improve Stability and Efficiency in Continual Learning

Joint or Disjoint: Mixing Training Regimes for Early-Exit Models

GUIDE: Guidance-based Incremental Learning with Diffusion Models