Patrick Putzky — accepted AI-conference papers

2026

Float8@2bits: Entropy Coding Enables Data-Free Model Compression

Patrick Putzky, Martin Genzel, Mattes Mollenhauer, Sebastian Schulze, Thomas Wollmann, Stefan Dietzel

ICML 2026poster

Post-training compression is currently divided into two contrasting regimes. On the one hand, fast, data-free, and model-agnostic methods (e.g., NF4 or HQQ) offer maximum accessibility but suffer from functional collapse at extreme bit-rates below 4 bits. On the other hand, techniques leveraging cal…

Cited by 0 Source ↗Scholar ↗

2026

The Loss Is Not Enough: Sampling Conditions and Inductive Bias in Contrastive Representation Learning

Justinas Zaliaduonis, Sergios Gatidis, Till Richter, Patrick Putzky

ICML 2026poster

Contrastive learning has emerged as a powerful paradigm for self-supervised representation learning, yet the precise conditions under which it recovers meaningful latent structure remain incompletely understood. We develop a measure-theoretic framework that formalizes the diversity condition, a requ…

Cited by 0 Source ↗Scholar ↗

2019

Invert to Learn to Invert

Patrick Putzky, Max Welling

NeurIPS 2019poster

Iterative learning to infer approaches have become popular solvers for inverse problems. However, their memory requirements during training grow linearly with model depth, limiting in practice model expressiveness. In this work, we propose an iterative inverse model with constant memory that relies…

Cited by 93 Source ↗PDF ↗code ↗Scholar ↗